// transcript — 8261 segments
0:00 What We’re Covering
0:01 Two years ago, I taught myself how to build AI agents without any prior
0:05 experience in AI. And since then, I've started multiple AI businesses that have
0:10 generated over $5 million in revenue. I've grown from 0 to 450,000 subscribers
0:15 here on YouTube. And I've built AI agents for some of the biggest companies
0:18 in the world. It's pretty safe to say that learning how to build AI agents has
0:22 completely changed my life. So, in this full course, I'm going to teach you
0:25 everything, and I mean everything that I have learned over the past 2 years about
0:29 building and more importantly making money with AI agents, even if you don't
0:33 know how to code. My hope is that you too can learn and use this incredibly
0:36 powerful skill in order to build the life of your dreams before these AI
0:40 agents start taking our jobs. As you can see, by the length of this video, I'm
0:43 not going to be holding anything back. So, in order to make it a bit easier to
0:46 consume, we're going to be splitting it into three different chapters. First,
0:49 we'll build your foundational understanding of AI agents, covering
0:52 what they are, how they work under the hood, and the key concepts that you need
0:56 to know before you actually start building them. And this is without any
0:59 technical background required. Secondly, we will be diving into four different
1:04 endto-end AI agent tutorials, taking you over my shoulder every step of the way
1:07 as we build some of the most popular AI agent use cases on the market today. And
1:11 I have carefully planned these builds out to give you a taste of multiple
1:14 different noode platforms and different AI agent types. and you'll learn how to
1:18 build each of these step by step watching over my shoulder. And finally,
1:21 I will give you my proven blueprint for monetizing your AI agent building skills
1:25 over the coming years while this technology continues to explode in
1:28 adoption and popularity. I'll be sharing the exact strategies that I've used to
1:31 generate millions of dollars with this skill set over the past 2 years. So,
1:35 let's get into it. Now, if you're new to the channel, let me quickly share why
1:38 I'm qualified to teach you about AI agents. So, if you are new to the
1:41 channel and don't know who I am, my name is Liam Mley and 2 years ago, I started
1:45 learning about AI with no prior experience in the field. I was teaching
1:48 myself how to build AI agents and chat bots all through my own self-study,
1:52 which I documented right here on this YouTube channel from day one. So, you
1:54 can go back and watch all my previous videos of how I got from there to here.
1:58 This led me to starting Morningside AI, which is my AI automation agency where
2:02 we build AI systems and agents for businesses from basic customer support
2:06 systems to full AI SAS platforms and even we've built my own AI agent SAS
2:10 platform, Agent, which is now over 45,000 users. At Morningside, we've
2:13 worked with publicly traded companies and even recently an MBA team. I also
2:17 run the world's largest AI business community with over 120,000 members on
2:21 school as of this recording. and through my community and through this YouTube
2:24 channel here, I've taught hundreds of thousands of people from all backgrounds
2:28 really how to build and make money from AI agents. So, everything that I'm about
2:32 to teach you today is exactly what helped me to achieve all of this. So,
2:41 in. Now, if you look at how long this video is, you'll realize that there is a
2:44 lot to cover here. And now, I don't want you to give up halfway through the
2:47 video. So, let's quickly take a moment just to get clear on why AI agents are
2:51 quite literally the next big thing. And I know that sounds cliche and you hear
2:54 it all the time, but seriously, they are. And why learning to build them is
2:58 by far one of the most valuable skills that anyone can have over the coming
3:02 decades. And if if my experience over the past 2 years is anything to go by,
3:05 that should be enough proof for you guys to believe me. So, stick with me. But
3:08 here's the hard truth about AI and jobs right now. According to the latest
3:12 research, McKenzie predicts that AI and agents could automate up to 50% of
3:16 current work by 2030. And the World Economic Forum reports that 41% of
3:20 companies plan to reduce their workforce due to AI. Now, a lot of this sounds
3:23 doom and gloom, and of course, many people are naturally worried about their
3:26 career in future based off seeing this kind of data. But it's not all bad if
3:30 you know where to look. And this video isn't about making you feel all sad.
3:33 It's about uplifting you. And if you look on the flip side of the same data,
3:37 these same reports reveal an enormous opportunity for those willing to seize
3:40 it. So the World Economic Forum's future of job report states that 50% of
3:44 employees plan to reorient their business in response to artificial
3:47 intelligence. And due to this reorientation, 66% of employees plan to
3:53 hire talent with specific AI skills such as prompt engineering. So on one hand,
3:57 we have the expectation of massive layoffs and automation of work over the
4:01 next 5 to 10 years. But on the other, we have the majority of employers searching
4:05 for people who have these AI skills or really just basic AI literacy. Why?
4:10 Because AI literate employees who can automate parts of their work can have 5
4:14 to 10x the output of someone who doesn't have any AI literacy. And I promise you
4:17 that brushing up on your AI and reaching this point of AI literacy so that you
4:21 can be on the winning side of this next 5 to 10 years is actually so much easier
4:25 than you think. I mean, it's easy as watching this entire video in order to
4:28 build your AI skills base to make a big step towards AI literacy. And if you
4:31 don't believe me when I say that a little bit of self-study like this video
4:35 goes a long way, here's a great clip that I've seen recently on the All-In
4:39 podcast from one of the most respected investors and technologists in the
4:42 world, Navar Raakan, alongside a whole bunch of other very smart billionaires.
4:46 Again, I would say the easiest way to see that AI is not taking jobs or
4:51 creating opportunities is go brush up on your AI. Learn a little bit, watch a few
4:55 videos, use the AI, tinker with it, and then go reapply for that job that
4:59 rejected you and watch how they pull you in. And so this video is exactly what
5:02 Naval is talking about. So whether you're an aspiring entrepreneur wanting
5:05 to learn valuable AI skills and launch an AI business like I have, or you're a
5:09 business owner wanting to just understand agents so you can use them to
5:13 grow your business, or maybe you're just wanting to make sure that you are the
5:16 last person that your boss thinks of firing because he or she's an AI wiz and
5:21 I can't afford to lose them. Then I have made this video for you guys. Now, what
5:23 I want you to do is close out all of your other tabs. Go get a notebook and a
5:27 pen and a beverage of choice and make sure you make a commitment to yourself
5:31 right now in order to finish this training and ensure that you are going
5:35 to be empowered by AI and not replaced by it. Now, if you've done all that,
5:39 let's get stuck into it. All right, so step one in building AI agents is knowing what the hell an
5:45 agent actually is. So, 2 years ago, when I first started learning about AI
5:49 agents, I had no idea what they actually were. The term AI agent gets thrown
5:52 around a lot in almost like everywhere these days. You got AI agents this, AI
5:57 agents that. But what actually is an AI agent? Well, the clearest definition
5:59 that I found that helps beginners to really wrap their head around what they
6:03 are is this. An AI agent is a digital worker that can understand instructions
6:08 and take actions in order to complete tasks. So, in a very simple way, just
6:12 like businesses have employees who handle different tasks, an AI agent is
6:17 like having a digital employee. But the cool thing is that you can build them
6:20 and you can make them do whatever you want. You're like literally building an
6:23 employee that you can put to work to do things for you. And of course, they cost
6:27 much less to run than a human and they don't need sick days and they don't
6:31 start beef with Mike over in the sales department because of his comment at the
6:34 coffee machine. So, I'm sure you can see the appeal of this kind of digital work
6:38 and AI agents to businesses who are looking to adopt them. In order to really understand why these
6:45 AI agents are such a big deal, we need to look at where we are coming from. So
6:48 most of you have probably encountered those chat bots on websites before. You
6:52 know those little little chat widgets that pop up saying like, "Hey, how can I
6:55 help you?" So these kinds of chat bots are pretty basic, right? They a lot of
6:58 the time they're useless and they're they're kind of like a waiter who can
7:02 only really recite the menu but can't actually take your order or or bring
7:05 your food. They can't do anything. They just respond with some kind of
7:08 pre-written answers. Well, nowadays it's a simple AI generated answer. But AI
7:13 agents are different, right? So, here's an example. If you ask a regular chatbot
7:17 about booking an appointment, it might say, "Oh, our business hours are 9 to5.
7:21 Please call to book." And that's it. They just give you some information
7:25 back. But with an AI agent, it could actually go and check the calendar, find
7:29 some available slots, go back and forth with the person that they're chatting to
7:32 in order to book an appointment, send you a confirmation email, then update
7:36 the business's scheduling system and CRM automatically in seconds. This ability
7:41 to take action is what makes agents so powerful. They're not just fancy chat
7:44 bots. They're actually digital workers who can search through databases, update
7:49 spreadsheets, send emails, book appointments, generate hold documents,
7:54 and much much more. And so building and deploying an AI agent is a bit like
7:58 hiring a new employee because when you bring someone into a business, you need
8:02 to firstly explain their roles and the responsibilities to them. You need to
8:06 give them access to your system so they can use them. And you need to trust them
8:10 to handle those tasks independently. And now when we are building agents, as we
8:14 see later, it's exactly the same, except these agents are going to be working
8:17 24/7. They're never going to get tired. They can be duplicated and modified
8:21 instantly. And they cost a fraction of what a human employee does. And this is
8:25 exactly why understanding how to build and sell AI agents is becoming such a
8:29 crucial and valuable skill these days. Because whether you're an entrepreneur
8:33 looking to scale your business or you're an employee wanting to become
8:36 irreplaceable and and make more money at work, knowing how to create and deploy
8:41 these digital workers is like the biggest cheat code in the whole world
8:44 right now. Now that you understand what AI agents actually are, let's look under
8:50 the hood and see how they actually work. Just like humans need a brain, memory,
8:55 and tools in order to do their job, AI agents need specific components in order
8:59 to function correctly. An AI agent needs five key parts in order to work.
9:03 Firstly, every AI agent needs a brain. In the AI world, we call this a large
9:07 language model or an LLM for short. And you've probably heard of some of these.
9:11 You've got GPT from OpenAI, Claude from Anthropic, Gemini from Google, etc. You
9:15 can think of the LLM as having a super smart intern who can understand your
9:19 instructions in plain English, and then figure out how to get things done from
9:22 those instructions. So, without this brain, all of the other parts would be
9:25 useless, right? It's like having a whole desk full of office supplies but having
9:28 no one sitting there in order to use them. Secondly, the brain needs
9:32 instructions on how to behave. And this is prompting. So writing a prompt for an
9:36 agent is how you program a lot of the behavior of it rather than having to
9:39 code it manually. And this is really what makes building AI agents so much
9:43 more accessible to non-coders as the way of actually programming the
9:46 functionality and how they work is done through clearly written instructions
9:49 rather than having to actually code it. Thirdly, agents need memory. Imagine
9:53 trying to have a conversation with someone who forgets everything you said
9:56 30 seconds ago, right? So, memory is really important because it allows your
10:00 agent to remember what you talked about just a few messages ago, keep track of
10:04 the tasks that it's been working on, build on previous conversations, and
10:08 even in more advanced ones, it can learn from your past interactions. And the
10:11 good news about memory is that most AI agent platforms completely handle this
10:14 memory component automatically. So, you don't need to worry too much about it.
10:17 But just know that it is an important part of a functioning AI agent. The
10:20 fourth component of an agent, and this one is optional, but it is external
10:25 knowledge. AI models like GPT and Gemini are pre-trained on a huge amount of
10:29 data, but that data is basically cut off at a certain point, eg 2024. It's kind
10:34 of like having a new employee who only really knows what they learned in
10:37 school. But just like you can train an employee like that with your company's
10:41 specific materials, you can also give an AI agent additional knowledge on top of
10:45 the information it was trained on through providing things like PDFs of
10:48 your company documents, spreadsheets with product information, customer
10:52 service transcript, or basically any other textbased information. Without
10:56 this added knowledge, agents will be limited to general information and
10:59 couldn't handle specific business tasks. But as I said, knowledge is optional and
11:02 you will only need it in some builds. Finally, and this is the most important
11:07 part, we have tools. So tools are what transform an AI agent from just being
11:11 able to chat to being able to actually get things done. So you can think of
11:14 tools like giving your digital employee access to different softwares. Just like
11:18 you might give a new hire access to your email or your calendar or your CRM
11:22 system, you can give an AI agent access to digital tools that let it take
11:26 actions when needed. These tools let your agent do things like checking
11:29 real-time data, updating databases, sending messages and notifications,
11:32 creating documents, all the stuff we went over just before and much, much
11:35 more. The really powerful part, which we're going to cover later, is when
11:38 agents use multiple tools together in order to solve complex problems, just
11:42 like us humans would use multiple different websites and softwares when
11:46 doing our tasks. Now, let me show you how all of these parts work together in
11:50 a real example. So, say you want an agent to handle customer support. When
11:54 the agent is sent a message, the brain immediately understands the prompt that
11:57 it has been given and also understands what the customer is asking. It checks
12:01 its recent memory before replying each time to understand the full context of
12:04 their conversation. And if the brand detects that the customer wants a
12:08 specific question answered from the knowledge base, it will use its external
12:12 knowledge in order to deliver the right information to them. And finally, it may
12:16 use tools to update a customer's account or to process a refund whenever required
12:20 during the conversation. So all of these things are happening in seconds as the
12:23 conversation is going on. Which is why AI agents are such a game changer. They
12:27 can combine all of these components in order to create a fully capable digital
12:31 worker that very very closely replicates how humans work. Now that you know the anatomy of
12:34 The Three Ingredients
12:38 an agent and the five parts of it, a more practical framework for
12:41 understanding how we actually plan and build AI agents is what I call the three
12:45 ingredients. Basically, you only have three elements to plan when creating an
12:49 AI agent, which when mixed in various ways can create millions and different
12:53 types of agents for different use cases. This is because the AI model or brain
12:57 can be easily swapped in and out and isn't really a major factor in the
13:01 performance of the agent as any of the top models that you pick from any of the
13:04 different providers at any given time, they're all pretty good. And also, the
13:07 recent chat memory is handled by default in almost all cases when you're building
13:09 on these platforms that you're going to see later. What this leaves us with is
13:12 what really matters when building and planning AI agents. Firstly, the
13:15 knowledge, the external data that you want the agent to be able to use when
13:19 answering. Secondly, the tools, the different actions that you want the
13:22 agent to be able to take, eg saving the contact info to the CRM or getting some
13:27 live data on stocks or sending an email. And then finally, prompting, which is
13:31 the glue that ties everything together and determines how the agent behaves.
13:35 So, write these down. While the agent has five components, the brain, the
13:38 prompt, the memory, the knowledge, and tools, your main focus as an AI agent
13:42 builder is in the three ingredients of prompting, knowledge, and tools. In the
13:45 next chapter, we'll be looking at how you actually build an agent using the
13:48 different combination of these three ingredients. But first, we need to dive
13:52 deeper into the keystone of understanding how to build your own
13:55 valuable digital workers. And it all comes down to tools. Now, we need to dig a lot deeper
13:58 The Web, APIS, and Tools Explained
14:02 on tools as they are by far the most powerful part of AI. agents. But in
14:06 order to understand deeply and be able to build powerful agents with them, we
14:10 need to take a few steps back and actually cover the basics of how
14:14 software and the web and internet as a whole works. Now, this is as techy as
14:17 it's going to get in this video, but I promise once you understand this, it's
14:20 so important and it's literally like having a superpower. So, please stick
14:24 with me through this. So, remember how we said that tools are what allow agents
14:28 to take an action to actually do things rather than just chat? Well, the way
14:33 agents use tools and do work online is just how we do it as well, but with one
14:37 key difference. Instead of clicking buttons and typing into forms, agents
14:42 use what we call APIs. And every time you use the internet, you're actually
14:46 making dozens of requests to APIs as well and getting responses back, but you
14:49 just don't realize it. So, let me show you what I mean. So, when you click on
14:53 this video, here's what actually happened. Firstly, your browser sent a
14:59 request to YouTube servers saying, "Hey, I want to watch this video." And then
15:02 YouTube servers sent back all of the data needed. And thirdly, your browser
15:07 unpacked that data and started playing the video on your screen. So this
15:10 request and response pattern happens with almost everything that you do
15:13 online. When you open up Instagram, you are requesting your feed from Instagram
15:17 service. When you send a tweet, you are sending your data through Twitter's
15:20 service. And when you check your email, you are requesting from Google the
15:24 latest messages in your inbox and they're sending it back and your browser
15:27 is loading it. Thankfully, we get pretty websites and apps that make it very easy
15:31 for us to do this and use software via APIs through a nice application. But
15:34 under the hood, it is still two computers talking back and forth,
15:38 requesting, sending, and displaying new information for us on our screen. These
15:41 request and response happen through what we call APIs, which are application
15:45 programming interfaces. So, you can think of APIs like waiters in a
15:48 restaurant. Basically, they're going to take your order or your request to the
15:52 kitchen, which are the servers of the business, and then they bring back your
15:56 food, which is the response. So, you have request and response, and you have
15:59 you as the client, and them as the server. There are two main types of
16:02 requests that you can make. Firstly, either a get request. This is basically
16:07 just like asking for information like checking the weather or looking up the
16:10 price or loading this video. You're requesting to get the information to do
16:14 something. Secondly, we have post requests, which is when you're sending
16:17 some kind of information like posting a tweet, sending an email, or uploading a
16:21 photo. So, go back and write both those down because we're going to be using
16:24 them extensively in the building section of this video. Now, here's where it gets
16:28 interesting. So, AI agents use these same APIs as their buttons to do things.
16:32 So, each tool an agent has access to use is essentially an API that it is able to
16:36 call. So, these kinds of tools come in two different flavors. We have pre-made
16:40 integrations like Google Calendar or Gmail where it kind of comes out of the
16:43 box ready for you to use and just plug straight into your agent. And then we
16:47 have custommade tools that we can build ourselves. So you can think of pre-made
16:50 integrations like buying a readymade meal where they've done a lot of the
16:54 hard work versus custom tools where we are like cooking from scratch. And both
16:58 work, but custom tools give you a lot more control. And this is a skill that
17:01 I'm going to be teaching you in the second chapter of this video. Okay. So now you got the basics.
17:04 Anatomy of a Tool
17:07 Let's get clear on how a tool is actually made and what the key parts are
17:11 as you're going to be using them a lot. So, let's break this down using a simple
17:15 example of a text capitalization tool. It takes in some text and the outputs
17:19 the capitalized version of it. So, first to create a tool, we need a function. We
17:23 need something that does work. In this case, it's super simple, right? It needs
17:26 to take in text and it needs to make it uppercase. So, this can either be done
17:30 through a basic Python function or you can use an LLM to do this as well.
17:33 Basically, we need to build some way to capitalize the text that we give to this
17:37 function and actually do the do the work. Next, in order for the AI agent to
17:41 use this function, we need to wrap it in an API. So, we have the function and
17:44 then the API wraps around it. And this is essentially making that functionality
17:49 we created accessible over the internet via APIs. Without it, the function
17:53 cannot be used by our agent. And in order to use the API that we've just
17:56 created and use this function inside it, the API is going to expect the same sort
18:01 of inputs that the function needs. So the input of the text that we want to
18:04 capitalize and it's going to output the capitalized version. So this is very
18:08 important to remember function. It takes in the input of the uncized text does
18:12 work and outputs the capitalized version. We're basically then just
18:16 building an API around it so that we can put it on the internet and then we can
18:19 have an agent that knows how to call that API can send information into the
18:23 input go through the function and then get spit out and then our agent catches
18:27 it at the end. But the magic step and what has really caused the AI revolution
18:31 to kick off is that we can explain to our agent how to use this API just by
18:36 explaining how the API works in natural language. And this is where schemas come
18:40 Schemas: API Instruction Manuals
18:42 in. A schema is like a one-page instruction manual on how to use an API
18:47 and therefore how to access the functionality inside that API. And when
18:50 an AI agent is given one of these schemas, it too can read that
18:54 instruction manual and determine things like what the tool does, what
18:58 information it needs as an input, like we talked about before, and what
19:02 information to expect as an output. Now, they may look scary, but they're
19:04 actually really, really easy to understand, and we're going to cover
19:06 them in the next chapter with this video. And the good part about it is
19:09 that these days, schemas are automatically created by many of these
19:12 no code platforms that you build agents on. But I'm teaching you this because it
19:16 still helps to know what they are doing and what that what's really happening
19:19 under the hood on these platforms. And there are still going to be times where
19:21 you may need to roll up your sleeves and do it yourself. The incredible part
19:24 about these schemas is that modern AI like chatpt can read these instructions
19:28 and perfectly understand not just how to use it and like okay I need an input and
19:32 then I expect an output but also when to use it. For example, let's say we had an
19:36 agent and we gave it that capitalization tool that we just talked about and then
19:39 we said can you please capitalize this text? Mary had a little lamb. The agent
19:43 would then read over the schemas that we provided it and then it would see that
19:46 there's a tool with a description saying this tool capitalizes text right in the
19:50 instructions for the capitalization tool. We would have said this thing is
19:54 for capitalizing text and it takes in some text and it gives you the
19:57 capitalized version. And so the agent will read that and see okay this looks
20:00 like based off the instruction they just gave this is the tool that they want to
20:03 use. And then it will check the requirements and see that the tool takes
20:07 in one input in string format which is just text which we have described as the
20:11 text to be capitalized. So it reads all this. He says okay it it needs one
20:15 input. It's in string format. So I know I need to give it some text and okay
20:18 what does this text do? It's the text that they want to capitalize. Great. So
20:22 now it knows it needs the input and it knows that this is where it's going to
20:24 send the text to be capitalized. Then now that it knows what it wants, it goes
20:28 back to our message and it intelligently extracts Mary had a little lamp. not,
20:32 hey, can you please capitalize Mary had a little lamb? It's smart enough to know
20:35 that we want that taken out. So, it will take that part, Mary had a little lamb,
20:39 out of our input, and then it sends that to the API where our capitalization
20:43 function does its thing. Then the API sends back the capitalized version plus
20:48 a bunch of other response data as well. Then the agent looks at your original
20:52 question, looks at this messy response it got back from the API, and then using
20:56 its brain, the LLM, it writes a natural language response answering your
20:59 question. It would say, "Here's your capitalized text colon Mary had a little
21:03 lamb in all caps." That may sound complicated. It may have gone over your
21:06 head. Please go back and just listen to it again. You really, really need to
21:10 understand this process of uh the message comes in, looks at the schema,
21:13 realizes, okay, it wants to use this tool. Okay. What do I need to do in
21:16 order to use this tool? Okay. Well, then I'm going to grab it out of the input.
21:19 I'm going to put it in here. And it can actually go back and forth. Say our
21:22 capitalization tool needed some other input. Say you needed to provide uh the
21:26 number of letters you wanted to be capitalized. It may see that this tool
21:30 needs two inputs and I've only been given one. So then it will go back and
21:34 ask me, hey, could you can you please tell me how many letters you want to be
21:37 capitalized and you will see this magic in the agents that we're going to build.
21:40 When the agent can ask you questions in order to help fulfill the needs of the
21:43 tool, you have this very intelligent system that really will blow you away
21:46 when you see it in action. And one thing many people miss about this process is
21:50 the agent actually gets back raw computer data from the API or what we
21:55 call JSON. But using the LLM, it can transform that into natural conversation
21:59 and answer your question in a very very uh clear and concise way. So it's
22:02 basically like having an employee who can read all this technical information
22:05 and then explain it to you in plain English, which is another part of why AI
22:09 agents are so powerful. And so when you understand this pattern that we've just
22:11 gone through, I promise you, you will never see the internet the same way
22:15 again. Every action online is just requests and responses. And therefore,
22:20 we can build our own tools and AI agents to automate all of it. So instead of you
22:24 manually searching the web, copying information, pasting it into
22:28 spreadsheets, sending emails, an AI agent can do it all automatically using
22:31 tools if you build it correctly. It's like having a digital employee who can
22:36 press all of these API buttons for you thousands of times faster than any human
22:39 could. And don't worry if this feels a little bit technical. In the next
22:43 chapter, uh I'm going to show you how to create your own tools like this from
22:46 scratch using platforms like Relevance AI, uh where you can build out powerful
22:49 tools without writing any code. and will really start to click into place once
22:52 you see the stuff in action in the building section. But before we get into
22:55 that, let me reveal the power of AI agents which is unleashed when they are
23:00 given multiple tools to work with. Now, obviously having an AI agent
23:04 that just capitalizes text isn't very useful. I get that. The real magic
23:08 happens when you give agents multiple tools and the ability to use them
23:11 together in order to achieve complex goals. So, do you remember our
23:15 definition? AI agents are workers that can understand instructions and take
23:19 actions to complete tasks. When you give an AI agent a task, it's going to try
23:23 its best to execute on it, but if it doesn't have the right tools on hand to
23:26 do the job, it's going to be useless. And so, the more tools that you can give
23:30 an agent, the more flexibility it has to solve problems just like a human would.
23:33 So, let me give you a real example from my own business, right? Say I build an
23:37 agent and give it the task. Find AI startups that have recently raised money
23:41 and put them in a spreadsheet and add a summary of each of the businesses in the
23:44 spreadsheet and then email me the link to the spreadsheet. When you give an AI
23:47 agent a task like this and provide it with multiple tools to use, it can break
23:51 down this problem just like a human would. For example, it might think first
23:56 I need to search for AI startups using my web searching tool. Okay, let's do
23:59 that first. Then I'll need to create a new spreadsheet with my Google Sheets
24:03 tool. And then for each company that I find, I'll need to add a row to the
24:06 spreadsheet. And then I'll need to write a summary of each business and put it in
24:09 a new column. And then finally, I'll use my email tools in order to send the link
24:12 to Liam. And that's all great, but then when you add on top of that powerful
24:16 reasoning models like OpenAI's 01 and 03 and even things like deepseat as the
24:20 brain of the agent that can plan, take actions, then reflect and then plan
24:24 again and so on. You have essentially created a truly intelligent AI that
24:28 solves problems and approaches them just like a human would. So, say for example,
24:32 the original plan was to use the web search tool to search for AI startups
24:35 raising money. Probably a terrible search term, but what if that doesn't
24:39 return any good results to the agent? Well, a human would go, damn, I need to
24:43 change my search term or maybe I need to try find a different method of finding
24:46 these companies on like LinkedIn or something. The latest in AI technology
24:49 like these reasoning models, it allows these agents to do this exact same kind
24:54 of reflection and replanning in order to achieve their objective. And this is
24:57 when you can really see why we call them digital workers because they can do
25:01 things like planning multiple steps. They will use different tools in a
25:05 sequence and even adjust their approach based on the results from those tools.
25:08 Now, I should mention that this technology isn't perfect yet, right? So,
25:12 these multi-step tasks are often unreliable and agents typically need
25:16 human supervision for more complex workflows. But things are moving
25:21 incredibly, incredibly fast. In fact, we're already seeing the next evolution,
25:24 which is multiple agents working together. Instead of just one agent
25:27 trying to do everything, you can have one main agent that you give orders to,
25:30 and then it can use all of the other agents underneath it as tools where it
25:34 can send specific instructions. Like underneath the main agent might be a
25:38 research agent, which is best at finding companies and has its own tools. Then
25:41 you have a writing agent that's really good at writing summaries. Then you have
25:44 an emailing agent, which has got all the emailing tools. And so each of these
25:47 agents can be specialized in their specific task with multiple tools. and
25:50 then they all work together to achieve a common goal. This is exactly what major
25:55 companies like HubSpot and Microsoft and Google are building towards. It's these
25:59 entire workforces of AI agents that can handle complex business processes
26:02 automatically. In the next chapter, I'll show you how to build AI agents like
26:05 this for yourself using no code tools. But first, we need to understand the
26:08 different ways that these agents can actually be used in the real
26:14 world. So, we understand how AI agents and tools work under the hood. Now,
26:18 great. If you don't, please go back and take some notes, right? You should by
26:21 now have a whole bunch of notes um from the stuff that we've covered already.
26:23 And this stuff that you're learning took me two years in order to learn and and
26:27 be able to apply effectively. So, you best believe it that it's going to take
26:30 you two to three watches before it all sinks in. So, if you're feeling a bit
26:33 lost and and overwhelmed, don't worry. That's how it feels with learning
26:36 anything new or how it should feel if you're learning something that's
26:38 actually pushing your boundaries and adding something to your to your
26:41 capabilities. Next, we need to look at the different ways that AI agents can be
26:45 used in the real world. There are two main categories of AI agents.
26:49 Conversational agents and automated agents. Conversational agents are ones
26:52 that humans interact with directly through chat on things like websites.
26:56 You've got maybe you're chatting to it on WhatsApp. You've got interacting with
27:00 it over the phone via phone call. You've got chatting to it via Instagram DMs or
27:04 custom apps and websites. For example, OpenAI's GPT platforms allows you to
27:08 create agents that you can chat with directly on your computer or on your
27:11 phone. or using platforms like my own Agent, you can connect these agents that
27:15 you build onto a WhatsApp number or onto Instagram. And I'll show you how to do
27:19 this in the tutorial chapter of this video. So, in all these cases, you or
27:23 someone else is there sending messages or instructions to the agent and
27:26 explaining what you want to do and kind of chatting back and forth with it,
27:29 whether it's on a website, WhatsApp, Instagram, or whatever. And within these
27:32 conversational agents, it's not just text based. It's like I said, there's AI
27:35 voice agents as well, which are an extremely exciting sector of the AI
27:39 space right now. And these systems use multimodal models that can take in audio
27:44 as input and then produce audio as an output. And so these agents can be
27:48 chatted to over the phone or via audio rather than via text. This AI voice
27:52 stuff is super cool. And in the tutorial section, I'm going to show you how to
27:55 take the exact same AI agent that we can chat to on a website and then connect it
27:58 to a phone number and talk to it on the phone. But then we get to what I call
28:02 automated agents. And so these are slightly different from the
28:04 conversational ones. The truth is that AI agents don't always need humans to
28:09 talk to them and use them directly. All they need is some kind of input or
28:13 instructions to trigger them and that tells them what to do. This means that
28:16 we can build these automated agents that instead of waiting for some kind of
28:20 human input, they are actually part of larger systems and processes and they're
28:24 triggered automatically by events like a new email received or a form submission
28:28 or they work on schedules like once a day and they essentially work in the
28:32 background without necessarily having human oversight or input. For example,
28:35 later in the video, we are going to be building an automated agent that is
28:39 triggered by a new form submission. When the form is submitted, some of that form
28:42 data is taken and sent to the agent, which then causes it to use the tools
28:46 that we've equipped it with and follows the instructions in the prompt that we
28:49 gave it in order to make decisions and take appropriate actions on our behalf
28:53 in a fully automated way. We are still sending the message to the agent, but
28:57 it's not a human needing to type it manually or speak it over the phone.
29:01 There's no human step. The input is being automated in some way. And this of
29:04 course opens up a huge number of use cases for AI agents in businesses
29:07 especially. And of course I'll be showing you how to build both types of
29:10 these conversational and automated agents in the tutorial section of this
29:14 video. But the last step of building your foundation of knowledge before we
29:17 move into that is to look at some real world examples of how businesses are
29:23 using these AI agents right now. So firstly we have the personal
29:27 assistant category. And this is what most people think of when they hear the
29:30 word agent. something that you can chat to that's going to update your calendar
29:33 and sort of send emails and even make phone calls for you. Um, now these are
29:38 all nice to have features, but honestly uh this space is likely going to be
29:41 dominated by the big tech giants. You've got OpenAI through Chatbt trying to do
29:45 this with Tasks, Google through their suite of apps and connecting them to
29:49 Gemini and Apple through Siri. These guys are going to eat up this entire
29:52 market of personal assistance and your own personal AI agent that helps you do
29:56 personal stuff. the real opportunity lies in business applications and how
30:00 people like you and I can build and sell AI agents to businesses which we're
30:03 going to be covering in depth in the final chapter of this video. So, we've
30:05 got the next chapter which is going to be on building the four tutorials and
30:09 the final chapter is all about how to sell and how to monetize your AI agent
30:13 skills that you've just learned. One of the core use cases for businesses right
30:16 now are what's called co-pilots. And these are AI agents made for specific
30:20 roles in a business. We're going to be building one of these later in the
30:23 video. And these specialized AI agents are essentially helping someone in a
30:27 specific role in a business to do their job more effectively. Take a customer
30:31 support co-pilot for example. It would have a knowledge base that allows reps
30:34 to get answers to customer queries instantly and deliver them over the
30:36 phone. So they've got the little co-pilot up on the side there. They're
30:40 on the phone. They get a question, they can search and for an answer in the
30:42 knowledge base, it gives them back and they can give it to them over the phone.
30:45 This same agent could also have a tool that allows them to look up the customer
30:49 information very quickly. Um, I could have another tool that it makes it very
30:52 easy to send a summary of the call into the database so that the next rep who
30:55 picks up the phone and talks with them knows exactly what was discussed
30:58 previously. It's like giving every support rep some kind of AI assistant
31:01 that makes them dramatically more effective. It also makes their customer
31:05 support a lot more consistent as to what the company wants people to be saying,
31:08 which is a a big problem with managing large customer support systems. And then
31:11 we have lead generation and appointment setting agents. These are probably the
31:15 most valuable type right now. And businesses are using these on their
31:18 websites, through WhatsApp, on Instagram, and even over the phone to
31:21 engage and have conversations with the interested people who are approaching
31:25 the business 24/7. They can offer instant answers about products and
31:28 services. And they're even smart enough to be able to capture emails and phone
31:32 numbers mid-con conversation for later follow-up by sales team. Some can even
31:36 book appointments on the spot and mid- conversation by using a tool to check
31:39 the calendar availability and then using another tool to create a new booking
31:43 once they've agreed on a time with the prospect. Another real world agent use
31:46 case and one of my favorites is a research agent. And so these can help
31:50 businesses by automatically researching leads that come in through their website
31:53 or elsewhere. And when someone fills out a form, the agent can spring into action
31:56 and start searching the web for information on the company, finding
31:59 their LinkedIn profile of the person they're going to get on a call with and
32:02 gathering any other valuable data that it can find. Then it can take all of
32:05 this information and generate a summary of who this person is and what this
32:09 company is also and decide whether they're a good fit for working with the
32:13 company and if so then they can send the sales team some kind of detailed brief
32:17 or suggested strategy on how to close this particular person on a call based
32:20 on the research. So it's basically like having a an automated team of
32:23 researchers who as soon as leads show interest in your business, they're out
32:26 there figuring out everything about them and determining one whether they're a
32:29 good fit for you and your products and services which is called qualification.
32:32 Then secondly, if they are qualified, giving the sales rep something that will
32:35 bring them up to speed on who this person or who this company is and how
32:39 they can try to close them. So, we have covered a lot so far.
32:43 So, before we dive into each of these agent builds that I'm going to walk you
32:47 through over my shoulder, please make sure that you've got your notes taken
32:50 out and the core concepts of this video so far understood properly. You should
32:54 be clear on things like what is the definition of an AI agent? What are the
32:59 five parts of an agent? How is building an AI agent like being a chef? And how
33:02 many ingredients do you have to play with? What are the two main parts of a
33:08 tool? And what do schemers do? So, pause the video now and try to answer these
33:11 questions. And if you aren't 100% confident, you need to go back and watch
33:15 it again. So, don't rush this or you're going to feel way out of depth when we
33:18 get into the tutorials that we're going to be covering next. But if you are,
33:21 congratulations. You are one step closer to AI literacy and becoming a much more
33:26 valuable uh participant in this global economy. So before we get into the
33:28 second chapter, there's just three very quick things from me. Firstly, if you
33:32 are a business owner who wants to fast track to becoming an AI leader within
33:36 your industry, at my agency, Morningside AI, we offer everything from AI
33:39 education and upskilling programs for executives and staff to AI strategy and
33:44 roadmap consulting and of course AI development services as well. So we
33:47 would love to help you get ahead. So feel free to get in touch via our
33:50 website in the description below. And secondly, at Morningside, we are hiring
33:53 for all sorts of roles right now. So whether you want to build AI systems for
33:56 some of the world's biggest companies that we have as clients or to help
34:00 produce videos like these that are seen by millions of people or create
34:04 educational material for thousands of businesses. Uh we have roles for all
34:06 sorts of things right now. So you can apply using the link in the description.
34:09 And please, even if you're just vaguely interested, I really recommend you just
34:12 check out the link and see what roles we're hiring for. Uh you never know
34:16 what's going to be on there. Um and it may be a very good way for you to use
34:18 your skills to fast track into the AI space by working under myself and my
34:22 team. And finally, if you have gotten any value so far in this video, please
34:25 head down below and leave a like on the video. It helps me reach more people.
34:28 Um, I put a lot of work into these videos and it also lets me know that you
34:31 enjoy this kind of content and that I should make more of it. And of course,
34:33 if you like this kind of content and want to see more of it, you can
34:36 subscribe so that YouTube will put my videos up for you whenever a new one is
34:39 released. So, there's also a little share button if you want to click that.
34:41 That'll let YouTube know this is good content and that you're sharing it to
34:44 other people. Not only will that help me, but you can share it to your friends
34:48 and family who may also or you may want to help them to brush up on these skills
34:51 or help them give a way to get on the front foot with AI. And that's what I
34:54 really make these videos for. So, thank you for sitting through that little bit
34:57 of housekeeping and self-promotion. Now, building. I have carefully assembled
35:05 this chapter on building to give you the most bang for your buck possible in
35:09 order to kick off your AI agent learning journey. We are going to be covering
35:12 four different use cases across four different AI agent building platforms.
35:16 These are all no code, so don't worry about that. And the chances of you
35:19 falling in love with at least one of these platforms is pretty much 100% as
35:23 you're going to rapidly start to connect the dots uh about how you can start to
35:27 use these kinds of agents and these platforms in your own life or in your
35:30 work or for your friends and family and those around you. So, here's a quick
35:33 rundown of the builds we're going to be getting into. The first build is going
35:36 to be a sales co-pilot built with relevance AI. And here we're going to be
35:40 building three custom research tools from scratch, including an advanced web
35:44 scraping tool, which is a a great skill that I want to teach you. And with
35:46 these, we are going to be creating a conversational agent to help the sales
35:51 reps at Big Boy Recruits, a hypothetical fantasy uh recruitment firm, in order
35:55 for them to be better prepared for sales course. So that's the purpose of the
35:58 sales co-pilot. The second build is going to be an automated lead
36:01 qualification agent. And this will be built on a platform called N8N. And this
36:05 time we will be helping Big Boy Recruits, our fantasy recruitment firm,
36:09 to automatically research and qualify new leads and then send an email
36:12 notification to the correct sales rep. And this is going to show you that
36:15 automated style of agent where it's built into a process rather than having
36:18 a human input necessarily. In build number three, we will be building a
36:22 website and phone-based lead generation customer support agent. This will be
36:25 built on voice flow and it's going to be able to do three things. Firstly, answer
36:28 questions from a knowledge base, generate instant quotes using a custom
36:33 tool we build and also do lead capture on interested prospects. We're then
36:36 going to slap this agent onto a website widget so that you can chat to it via a
36:40 website and via chat and text. And then we're going to take that exact same
36:43 agent and connect them to a phone line so that we can call our agent over the
36:45 phone and access all of the same functionality we just talked about. And
36:49 finally, for build number four, we'll be using my own AI agent platform, Agent,
36:53 to rapidly build a lead generation agent and connect it to a WhatsApp number that
36:56 we can chat to. The leads that we collect are going to be automatically
36:59 sent into an Air Table database for later review. And please don't skip
37:02 around these builds as they're all kind of connected in some way where we're
37:05 reusing parts from build one and build two, etc. But without further ado, let's
37:09 get into building some agents. All right, people. Enough of the theory. Uh,
37:14 now we get into the fun bit of actually building these agents out. So, I've done
37:18 a lot of work and my team has done a lot of work. So, thank you to the my team
37:20 members who have helped me put this together. Um, putting together four
37:25 different AI agent builds for you. And this is really going to walk you through
37:28 an A to Z all the different platforms that you really need to care about, all
37:32 the different kind of core use cases and functionality. There's a lot more of
37:35 course, but this is going to really give you the foundation that you need to
37:39 succeed in the space. And hopefully it'll be the thing that kind of sparks
37:43 your interest in it because I I want you guys to have fun with it. these big
37:46 tutorials for me. Honestly, when I put a lot of work into it, I build up the sort
37:49 of mental resistance to it because I know how much work there is going into
37:52 it and I have to make this big whole session where I'm all uptight about it.
37:55 But I'm just going to try and relax and enjoy this. And I really want you all to
37:58 do the same. So, set a bit of time aside. You can either pause this video,
38:02 put on your watch later, but I really want you to take your time with this.
38:06 I'm going to be doing this more. So, when you do tutorials like this, there's
38:09 a few different ways you can do it. I can either do all the building and then
38:12 give you the templates and kind of just spoon feed it to you. And that's more so
38:15 what you do for someone if you're trying to like really fast track them and they
38:18 don't want to learn all the skills, but um I I know what I'm trying to build
38:21 here for you guys. And I'm going to give you a sort of stream of consciousness.
38:25 You just get to see me kind of jamming out and building these things. And I'll
38:28 be explaining my thought process and the concepts etc along the way to reinforce
38:33 what we've learned before. So I'm just going to dive into it with our first
38:40 And so what I've done is put together a big Figma board here which is going to
38:42 be breaking down all these different builds. So under here there's some
38:45 goodies you see. Oh, there's some goodies under each of these that I've
38:48 put together. Um and we're going to go through them one by one. Starting off
38:51 with agent one over here. I mean there's a lot of stuff here um that you guys are
38:54 going to get. So you'll get the whole Figma and it includes all the templates.
38:57 So if you do want to just kind of watch through this, pick it up. You can either
39:00 do it and follow it step by step with me and see how I build it and really build
39:03 those flexible skills that you're going to need to succeed in the space or you
39:07 can just watch it and be like, "Okay, I kind of get what he's doing and then
39:10 take all the templates from me at the end." That's I mean, completely up to
39:13 you. Depends if you want to be a really really nerdy builder about it and get
39:17 into the weed like like I like to do. Um or you just want to be like, "Hey, I
39:19 want to do this my business. I want to roughly understand how these things work
39:23 and what platforms." So, use this resource as you will. But we're going to
39:26 jump into agent build number one here, which is our sales co-pilot built with
39:30 relevance AI. So, running through this quickly, we have the purpose of this.
39:33 This is basically going to look a bit like this. It's going to be a co-pilot
39:37 and co-pilots work in that you have a uh it's basically a specific AI agent that
39:42 you build for a specific staff or staff member or role. So, say this case, it's
39:46 going to be a sales co-pilot. It'll be the thing that the sales rep uses to uh
39:50 in their day-to-day as they're working on their jobs. You can add tools like in
39:54 this case, you see we're going to have three different tools here for our
39:56 agent. One's going to be a company researcher tool. So this is when the
40:00 sales rep would be like, hey, I have a call coming up soon. Um, let's put in
40:03 this I need to research this company cuz this is who I'm going to be on a call
40:06 with. So they'll put in the company URL. This tool that we're going to create is
40:08 going to go and research that company. It's going to bring back and give a
40:11 summary. And then it's like, okay, well this is the LinkedIn URL of the person
40:14 that we're going to be got on a call with shortly. It's going to pass in the
40:17 LinkedIn URL. It's going to take that URL. It's going to pull all the
40:20 information and write a summary about the person. So now we have the company
40:23 summary and we have the person summary. And the final step here is going to be
40:26 what I'm calling a pre-all report generator. And that's going to take both
40:29 that company and prospect research that we've done. It's going to combine them
40:34 together and be for this specific company. As you're going to see that
40:37 this hypothetical company we're building this sales co-pilot for, it's going to
40:41 generate a basically a pre-core report or a strategy uh a strategy prep for the
40:44 sales rep so that they go onto those calls much more prepared and also sort
40:49 of a personalized guide on how to try to close this person. So, um, all of these
40:53 templates are going to be here. Each of these are templates for the tools. And
40:57 this is for the agent as well. Um, but here's some more information. You guys
40:59 can pick through this as you wish. But, um, this is the kind of end result and
41:02 we're going to be able to chat to it. And this would be something you could
41:04 build for a client. You could build it for your own business or you could just
41:07 tinker around. You could build co-pilots like this on relevance for yourself. So,
41:10 that's why I want to start with relevance because it also is a platform
41:13 that we can build these tools on. So, it's a really, really good one to start
41:16 with and let's get into it. So, the first step of course is to go to
41:19 relevance AI. So, I'll put a little link up here. You guys will be able to get
41:22 this Figma. It'll be on the school. Um, all of the information and all the
41:24 resources for this are going to be like this Figma is going to be linked to the
41:27 school. My free school community. If you haven't already joined, biggest AI
41:31 community on school, biggest AI business community probably in the whole world.
41:35 Um, so we can jump across that first link in the description. You'll be able
41:37 to find this in the YouTube resources section. Um, pretty straightforward. Of
41:41 course, when you click on this, it's going to ask you to log into relevance.
41:44 So, if you haven't already, you can make an account. Um, it's fairly low cost.
41:48 They have a free plan, then a team's plan, I believe. Um, so it's not too
41:50 much, but it is a really, really valuable tool as you're going to see.
41:54 So, you can sign in here. I'm going to jump in with my Google account. There
41:58 may be a bit of setup for your account, but I'm sure you guys are smart enough
42:01 to figure out how to set up an account. I'm sure relevance also makes it easy
42:04 enough. So, then we get taken to this dashboard, but we see on this left hand
42:07 side, we have tools. So these are the tools that I've talked about um where
42:10 it's some kind of functionality that we can create and we can build it all on
42:13 relevance no code and there's even sort of extensibility or you can add more
42:18 functionality in to relevance by adding some low code components or even custom
42:21 code. So relevance is a really great base for building not only tools but
42:24 then the agents that you can connect that into and we're going to use the
42:27 same relevance tools that we make now in multiple of the different agents that
42:29 I'm going to make for the rest of this video. So first things first if we go
42:32 back to the Figma here we see we need to make three different tools. So tool one
42:35 is company researcher. It's going to take in a company URL. It's going to
42:37 search the web and it's going to return a summary. So, that's the functionality
42:42 we need. Let's go and create a new tool. Going to call this um
42:48 research company. We can give it a cool little I'm going to zoom this up for you
42:51 guys. Hopefully, that's the right size. Um uh search. Have fun with this stuff,
42:57 guys. Like, if if it's putting stupid emojis on things to to enjoy it, then
43:01 then that's what you need to do. Like, uh if you make it a chore, it's going to
43:04 feel like a chore, right? Um, now we get to descriptions. Um, this is something
43:07 we're going to see recurring, but basically as as you know, as we learned
43:11 in the concept section, we need to have natural language descriptions of our
43:15 tools and of our APIs so that the agents can read those descriptions and
43:19 understand uh what the agent or what the tool does and what those different parts
43:22 does. So, you'll see this recurring throughout this. But first things first,
43:26 what does this research company tool do? Um, takes and oh, I got caps lock on.
43:32 takes um and there's also some tutorials here if you want to go deeper. Relevance
43:36 has some great documentation as well. Um but takes in a company
43:45 penny URL and scrapes the website then returns a sum and AI generated summary. So then to
43:57 build a tool we need to have some kind of input. You don't always need an
43:59 input. It can actually just be triggering, but generally you're going
44:02 to have some kind of input that the agent needs to pass into the tool. In
44:05 this case, to research company, it's going to need to take in some text,
44:09 which is going to be a URL. Um, we're going to say comp company
44:15 URL. And then again, here we have another description. You see, describe
44:18 how to fill this input. This is again going to help our agent within relevance
44:22 AI and elsewhere as you'll see in tutorial number four. Why this is so
44:25 important to add in the descriptions, right? So this is a URL for a company to be
44:39 researched must be in the format https colon slash um dot dot dot dot dot. So we need to
44:48 have the https for this to work. So that's going to be our input. Now if
44:50 this seems a bit confusing just stick with me. It will make sense in a second.
44:53 this stuff. If there's anything that I've learned from picking up so many
44:56 different tools, like when I first got into Facebook ads, when I've got into
44:59 building these kinds of agents, it's you feel completely overwhelmed, but that's
45:02 all just part of the process. And what feels difficult now is not going to feel
45:05 difficult forever. So, just please stick with me. Um, and it's a really, really
45:09 great feeling once you go and be like, this was hard a few weeks ago and now
45:12 it's really easy. So, we've got our first input. That's what the research
45:15 company tool is going to do. And then we need to define our steps. So, the next
45:19 step, we can go add. I'm going to hide this so we get a bit more space. Add
45:22 step. Now, the cool thing about relevance is that it comes with a lot of
45:25 great functionality out of the box. Here's one, extract website content, we
45:30 have LLMs, we have Google searching, we have all sorts of AI generations,
45:35 replicate um knowledge bases as well. There's so much cool stuff on here and
45:38 this is why I really rate relevance as one of the best platforms. If you were
45:42 to go all in and want to upskill, you can build so much on this. Um, so I'm a
45:45 big fan. I love the the relevance team and what they're doing. So the initial
45:48 plan for this build I was going to use the extract website content which is
45:52 fairly straightforward. We can say oh one other thing um the company URL we
45:56 are going to be using this company URL throughout our tool here. So we can
46:00 change this text to say something more descriptive. So company
46:06 u URL. You guys are going to got going to think like think and write like
46:09 coders now cuz you need to uh use some kind of syntax and use some kind of
46:13 variable naming convention. This is a a standard one or you can do things like
46:18 company URL camel case but I prefer this format as I'm I'm mainly a Python kind
46:21 of guy myself. Now that we have that named we can use it in these kinds of
46:25 fields. So here you can see pops up use inputs. So basically when this tool is
46:29 run it's going to take the inputs or the information we put in the inputs and
46:32 it's going to pass it to different steps and use them as we describe within the
46:35 within the builder here. So let's run morningings.ai. Um, and then we can
46:49 click uh run here. So, that's going to go to my website and scrape the
46:54 information off of it. There we go. So, it's got all of this, but you see it's
46:59 just pulling back the first page. Um, and this is why I actually I shuffled
47:02 this around. And I want to show you guys how to do something a bit more advanced.
47:04 I know this is supposed to be a beginner tutorial, but this is not really that
47:08 useful and I it's a very easy thing for me to just bump this up to a little bit
47:12 more valuable. Um, while relevance tool here is great, we can do better. So,
47:16 we're actually going to delete this. Um, you can use that step for all sorts of
47:20 other things, but I really like what's called fire crawl firecroll web scraper. This is a cool
47:29 app. Um, firecrawl.dev. Shout out to the guys at firecraw. Basically, if we then
47:33 go HTT Oh, I should just see if they can doti and now do a free scrape for us
47:42 here. So, this is just going to do the single URL just like we got in
47:46 relevance. But the difference here is if we then go to crawl, if you hover over
47:51 this, it's going to crawl a URL and all of its accessible subpages outputting
47:55 the content from each page. So, instead of just taking that front page, it's
47:57 actually going to crawl through multiple things. So this is really the first cool
48:01 thing or or first skill that I want to put in your tool belt is that you have
48:04 things like fire call that you can use their relevance. They have things like
48:06 map which is just going to output all the URLs that it finds. Then there's
48:10 other things here where you can use AI to extract data. I'm not going to go
48:12 into that. But what we want is this crawling functionality from firecraw. So
48:17 I'll put a link in on the school post that comes with this video. It again
48:19 first link in the description to go to the school and if you go to the YouTube
48:23 resources section um there will be a uh a whole post on this and all the
48:26 resources will be in there. It'll also be in my free course on school as well.
48:29 So you can find it in the classroom section. So you want to sign up to FCL
48:34 so you can get API key. It's very easy. We can just go through with Google.
48:37 Again, this is not sponsored and there is zero sponsoring going on through any
48:40 of these tools. I guess I'm kind of sponsoring my own tool because I'm
48:43 putting it at the end. But I'm not getting paid a dime for any of this. I'm
48:45 really just trying to put you guys on what I like to use, what's made me
48:49 money, what's made me a more valuable AI automation expert or developer for my
48:53 companies and for the companies you work with. continue and then you get to the
48:57 dashboard here. It might look a bit scary, but what you can do is go to the
49:00 API keys. So, you can click on create an API key here, YouTube. I'm having issues with mine. It
49:06 should be fine for you. I've already got an API key, but once you get the API
49:09 key, what you can do is take it and come back here to relevance. And you can go into the side
49:16 panel here. Where is it? Settings. And then we have our API keys. So, we're
49:19 going to need to add more into this later. So, keep an eye on this. This is
49:22 something you need to be familiar with. Um, and you can scroll down to firecrawl
49:25 and you just pop it into this firecall API key section here and you're good to
49:29 go. And you can come back. Oh no, we don't want to duplicate
49:35 that. Now we've got our firewall set up. We need to make a couple things. We need
49:39 to do a couple tweaks here. We have, of course, the if you want to get those
49:43 variables up, you can go bracket bracket um or curly bracket curly bracket, which
49:46 is shift um to get those. I don't know why it's not popping up. There we go.
49:51 Company URL. And if we hover over this, we can see it says scrape the provided
49:54 URL only. Uncheck if we want to crawl instead. So if we want to get that crawl
49:58 functionality that we just saw that we think we want to get all the data, we
50:02 can uh uncheck it. And then we want to extract the main content. So you might have to just trust
50:07 me on that one that we don't want to have all of the other rubbish. We just
50:11 want the body of the website. Um a number of pages. We don't want this to
50:13 take too long. You can expand this much more. Um but I'm just going to go for
50:18 say five for now just to keep it uh keep it quick. And then now what we can do is run this
50:24 again. We've still got my URL up here. We can go run step. Give it a second. You will at some
50:31 point have to pay firecrol. Of course, it's not a free service, but they do
50:33 have a free plan, so you should have no issues with getting that. So here you
50:35 can see we're getting a lot more data back from this web scrape than we were
50:39 with just the relevance version, which is great. So the next step is we have
50:42 this data. We want to generate some kind of research summary um so that we can
50:46 send that to our sales script once they have requested it. So now it gets into
50:51 the fun part of writing LLM prompts. For this one, sometimes you need to really
50:55 go and and make a big effort, which we are going to do later as you'll see. But
50:57 in this case, we just want a quick summary. I'm going to throw one in here
51:00 that I made earlier. All of this will be available on the Figma or it'll be given
51:03 somewhere in the resources, right? But um if we just need something quick and
51:06 dirty here, it's not really a massive part of the project, so it's okay to
51:09 just whack one in there. So bang, I've got it in there. Can you please take
51:12 this website content and summarize it into a 300word natural language summary,
51:15 which clearly outlines rad where they're based, their values, etc. anything that
51:19 would be helpful to know for a sales rep who will soon be on a call with them. Uh
51:22 break it into key areas like overview, products and services team etc. And I've
51:25 got a couple things here to make sure that it doesn't mess up which I it was
51:29 doing for me a bit in testing. So we do want to put in if we go curly bracket
51:32 curly bracket we want to put in the fire call data here which is going to be uh
51:36 all the website data that came back from our scrape. We want to then insert it
51:39 into this prompt here. So I hope you're starting to the things in your your cogs
51:42 and your brain are starting to click into gear here. And then we get to
51:45 select the model for this. It's pretty basic task. So, I want something quick
51:48 and cheap. Um, for many of you, it's going to be easiest to work with the
51:51 Open AI APIs because you've probably already played around with your API key
51:54 before, which I'll show you how to do in a second. But, let's go with GPT40 Mini,
51:58 but Relevance, of course, does have support for all of uh the other models.
52:02 But my tendency for most tasks now is to actually go for some of the Google
52:05 models that have come out. Again, you guys might be watching this in a year or
52:08 whatever. It might be quite different, but at the moment, Google is really
52:11 leading the way with making the cheapest models possible, and they're actually
52:14 really good as well. the the price decrease on using things like uh Google
52:17 Gemini Flash Light and Google Gemini Flash 2.0 and stuff like that. It's
52:21 ridiculously cheap and it's also a really good model. So, it's a bit more
52:24 difficult to get APIs on the Google side. So, I'm just going to stick with
52:27 OpenAI for this tutorial. Um, so let's go GPT4 mini. And then we need to of
52:31 course set up our API key. Um, if you scroll down, where is the OpenAI? So, to
52:39 platform.openai.com.playground/playgroundplayground, sorry. Um, this will again be linked in
52:43 the resources or you can just type up platform.openai.com. I'm sure you'll
52:46 find it. If you haven't already, create an account um, and sign in. And then you
52:50 need to go to your dashboard here. You go on the left side, you go to API keys,
52:54 and you click create a new secret key. And then you're going to be able to copy
52:57 that key and bring it back into uh, relevance. Paste it in here. And then
53:00 you're good to go and start using the OpenAI suite of models. It's pretty
53:04 easy, right? So now we have our information back from the scrape. We
53:09 have our prompt in here. And then we can go run step and see what this company
53:12 research tool is going to output for us. Boom. Morningside AI is a leading
53:15 artificial intelligence development company dedicated to empowering
53:18 businesses autonomous AI agent development, enterprise consulting,
53:23 chatbot development team. Uh they got keep me out of my own team page.
53:27 Damn. But uh yeah, so there you go. That's the that's the company research
53:31 tool. That's step one. Um I hope you guys can kind of see how that works. Now
53:33 a cool thing about relevance is there is this build section which we've just gone
53:36 through. But you can also go to use and this can be really helpful when sharing
53:39 these kinds of tools. So not this is not really only just an AI agent tutorial.
53:42 I'm also teaching you how to build tools because you can build very valuable
53:45 tools and something like relevance. And then you can go share. You can go
53:49 publicly available. Oh, and I can click this and then I can give to my employees. I can
53:56 give to the companies that I'm working with, the clients that I've sold these
53:59 kind of services on. And then they get a nice and handy tool like this. I mean, I
54:02 use these throughout my organization for like description generation, a lot of
54:06 content repurposing. There's tons and tons of different use cases for building
54:08 these kinds of tools. And then you can take this URL and you can share it
54:11 around to whoever you want. So, this is a uh a great way to use relevance. Let's
54:15 go back to our tool here. But for now, we need to keep moving along so we can
54:18 get this this first agent done. So, that's the first tool. I'm going to run
54:20 through it a bit quicker now that we know how these kind of tool buildings
54:24 work. I'm going to create a new tool here. I'm going to call it this time if
54:28 we go back to our Figma. And I recommend when you guys are building your agents,
54:30 you're building systems and planning them out. This kind of laying out in a
54:34 Figma, if you're not familiar, this is Figma. It's like a design software, but
54:36 you can also use it for kind of whiteboarding. It's called a Fig Jam
54:39 board. It's one of the types of uh boards you can do. And I use this all
54:43 the time with my team as well. It's like if you can't take what I'm telling you,
54:46 lay it out on a board so that I can review it and give you notes and and
54:50 then we all agree this is the build. Um there's often a lot of uh communication
54:53 issues with explaining functionality. So this AI agent layout, it's really
54:57 helpful. You can if it's maybe a workflow automation, you can do box box
55:01 arrow arrow arrow all of this laying out how it's going to be built. Here I've
55:04 done it for you in a very basic format. So we've done this first one. Maybe I
55:07 just make this green. Um second one is going to be prospect researcher. So this
55:11 takes in a LinkedIn URL, searches the web, and returns a summary. And you're
55:14 going to see how I tie this all into an agent shortly. So we're going to go research
55:22 prospect. Sil takes in a linked in URL scrapes the profile and then generates
55:28 an of the prospect input. We're going to need a link and URL the link linked and
55:35 URL of the prospect. this. Now, we're going to add a step and
55:46 relevance has got us here linked um get a LinkedIn profile or company
55:53 post. So, this is cool cuz then we can pop in our LinkedIn URL
56:03 them. So, we're going to get the user mine my LinkedIn profile, if you guys
56:12 want to connect with me on LinkedIn, more than welcome to do so. I'll put in
56:17 the description below. Um, we can do a little run step here. So, if we go back
56:22 over to data here, that's great. We've got my about section. So, this goes very
56:26 long way across because it's in uh it's in JSON here. It's got my company. It's
56:33 got my company domain, where I'm from, years, company, founded, tons and tons
56:36 of great information that you guys can use and we are going to use shortly. So,
56:39 this is really cool. I would probably add one more step to this if I was
56:42 taking this um and building it for for my own team. I would add in another
56:47 LinkedIn scrape here um where we just do the same thing, but we also get the
56:50 posts because posts can give you a bit more up to date um information on what
56:53 they've been doing recently. that you can guys can add that and you just go
56:57 add a step LinkedIn and you do the same thing as we've done here but you change
57:02 this to LinkedIn post get user post so that may be a cool thing for you guys to
57:05 add the uh functionality on at the end is a bit of a challenge you can pause
57:07 this video and do that and then we need a llm step to take this again I'm going
57:12 to grab a pre-written prompt that I did just to save some time going to drop
57:17 this in here says fairly similar stuff um LinkedIn data I'm going to put all
57:23 the data in there. And then we're going to use a GPT4 mini again. And we can give that a
57:30 run because we've already got this data queued up here. And there we go. We've
57:34 got a nice summary. Um, if I change this to nice and formatted. So there's a
57:38 little button down there between raw or formatted. And you can copy the stuff
57:41 out of here, of course. So Liam Mley, my followers. Damn, I got a lot more than I
57:44 thought. So there we have the summary, my name, where I'm based, uh,
57:48 information, my career experience. Super handy stuff. And this is going to be
57:51 super helpful in the next step when we generate that pre-core report. So very
57:55 quickly, we've created one more tool. I'm going to save this. So now we've
57:59 got, if we go back to our Figma, we have two of these done. Now the final one is
58:03 going to take in the company and prospect research and generate a
58:06 pre-call report. So this one's going to be a little bit different. If I go
58:11 create a new tool, preall report tool. Okay, free call takes in company
58:21 and prospect summary and generates a free call report for sales
58:26 direct. And now for the inputs for this, we need long text, not just normal text
58:29 because we're going to be taking in that big company and uh and prospect
58:34 research. So we go prospect summary summary of the prospect based on length
58:43 that and a prospect in this case is someone who's a potential customer. Just
58:47 to clarify that if you're new to business and don't really get these
58:58 summary, right? We have our prospect summary and company summary in there. I
59:01 hope you're following along. Next step is just an LLM step and we want to
59:04 combine these two together. Again, I've got a little handy prompt for this to
59:07 save us time. Now, in this case, you will see that the prompt is a bit
59:11 bigger, right? So this is um for more important parts when you are creating
59:14 tools whether it's it's for agents or just generally when you're using prompt
59:17 engineering and LLMs to create value. In this case we are creating value because
59:21 in this case we're taking in this prospect summary and this company
59:24 summary. We're also giving it the the context of this fantasy or or
59:28 hypothetical business that we are selling this agent to as like a co-pilot
59:32 system. We've got Big Boy Recruits which is Dallas based recruitment firm
59:35 specializing in software industry talent acquisition for SMBs. You're going to
59:38 see this kind of recur across the different projects we do, but basically
59:41 we're helping these big boy recruits to automate their business with AI. So,
59:45 this takes in some context on that business and it's going to generate a a
59:48 report that's going to help the sales rep say, "Okay, this is the company.
59:51 This is what we sell. This is what we specialize in. This is the company that
59:54 we're trying to sell to. This is who we're going to be talking to. What's
59:57 some how can I personalize this call or what's the strategy I can go into this
60:00 with? What are some angles that I can attack this call from?" And so to do
60:04 this, I have a prompt writing tool that I use quite regularly and my team uses
60:09 it as well. Um, perfect prompt. So, this tool does a lot of leg
60:13 work for myself and my team all the time. I'm going to give it to you guys
60:15 to use for free. You'll be able to clone it into your relevance account.
60:18 Basically, what I'll usually do is I'll put on uh the dictation thing. You've
60:22 got it on one of these keyboards. You've got this little thing. basically
60:26 whatever on your computer allows you to speak into the computer and it takes in
60:30 your voice and transcribes it into into text on the screen. I'll press that and
60:33 then I'll explain as you can see here what is this prompt doing and why and
60:36 I'll go this prompt needs to do this this this is going to take in this
60:39 information it's going to do this the reason we're doing this is this this and
60:43 I'll do like a big big body of text in there and then the next one if I have
60:46 them I'll give some good examples of input and output pairs of how I want it
60:49 to take in data and how I want it to spit it out. If you give it both of
60:52 these and you hit run, it's going to print you out using the researchbacked
60:55 prompting techniques that we use at Morningside. It's all crammed in here.
60:59 There's a video that I recommend all of you watch. It's going to be in the free
61:01 course anyway on school. So, when you get in there and watch my prompt
61:04 engineering guide, um, this is basically the entire information of that prompt
61:08 engineering guide smashed into this LLM step here. So, when you pass this
61:10 information in, it applies all of that and it gives you out a prompt that is
61:14 fully researched back and performs very, very well right out of the box. So,
61:16 that's a little bit of extra value I wanted to throw in there for you guys.
61:19 this is going to be available on the school um with the rest of the resources
61:22 as well. So basically I put in the information here about what this
61:26 particular uh task was. I said it's going to take in the prospect
61:28 information. It's going to take in the prospect summary and the company
61:31 information and it spat out this basically first go and I just had to
61:34 insert these variables. So this prompt and everything else will be on the on
61:37 the score resources as well. So in this case because we are doing a bit of
61:40 strategy and sort of high level thinking rather than just summarizing and we may
61:43 want to change the model here to something a bit smarter. We could go to
61:47 03 mini which is one of the later ones. Um, again, when you're watching this, it
61:50 might be 06 or 010 or whatever the hell they come up with next, but there's
61:53 probably going to be some much better models. So, just use a smart one because
61:57 it's really strategizing on how big boy recruits can position themselves for
62:00 this call. So, enough of me yapping about that. Let me grab some inputs for
62:06 [Music] it. Shout out muscle. All right. So, I've got this information here about
62:12 myself, my LinkedIn profile, and my company. And so again, remember that
62:15 this is for Big Boy Recruits, a Dallas recruitment firm specialized in software
62:18 industry talent. So it's going to look at my company, Morningside AI. It's
62:21 going to look at me and my background and my profile on LinkedIn. Then it's
62:25 going to spin, as we see here, um, review this, analyze this, map big boys
62:29 unique value proposition, ra, and it's going to try and create a report that's
62:33 going to allow the sales rep to sell me or close me better on there or find some
62:37 angles to sell to at least. So if we go run tool, and now you see that I'm using
62:40 a lot of just basic web scraping and LLM steps. I just want to show you guys the
62:43 basics. The thing is tools can get very very advanced when you have like CRM you
62:47 want to integrate into, but relevance allows you to do all of that. It's just
62:50 within the scope of tutorial, it can be pretty difficult to be pulling
62:52 information from all over the here cuz I have to set up a database, show you guys
62:56 how to do it, too. So, this this keeps it quite confined, but it still gives
62:59 you a good taste of it. So, if we look at this view, all key business
63:03 challenges and opportunity, Morningside AI, this Liam's profile gives me a bit
63:07 of a rundown of this mapping big boy recruits unique value proposition. Maybe
63:10 it's going to be better if I change this to the format. There we go. Talks about
63:15 mapping big boys recruits, strategic talking points. I've been following your
63:20 journey of digital marketing, AI, ra um dive into opportunity. I work at big
63:23 boys. This assist, and it's even gone and done a section on anticipated
63:26 objections. So, the idea is that the sales rep is going to have a skimmer of
63:29 this before the call, which ties into the value that I've listed here, which
63:32 I'm going to do for all of these builds, by the way, which comes down to
63:35 ultimately a better prepared sales rep should close more deals, right? if they
63:38 know more about the prospect and the company and you have an angle to try
63:42 sell through or suggestions at least. It should increase the conversion rate of
63:45 the sales team. So, we've built this tool. We can change this to green now.
63:48 And the final step is going to be heading over to our agent builder within
63:52 relevance. I'm going to save this. If you pop over in the left panel here, you
63:56 can go into our agents. And what we want to do is create a new
64:03 agent. We're going to call it our sales co-pilot. Um, big boy big boy sales co-pilot. Sorry, I
64:12 got, like I said, I got to have fun with the stuff where I go kind of insane. Um,
64:18 this this agent is our sales co-pilot that helps reps to be better prepared for sales
64:30 [Music] calls. Triggers, we don't need to do any of that. We go to core instructions. I'm
64:35 going to again paste in some of the stuff that I've prepped earlier. So, if
64:37 I paste this in here, you'll see that it's structured fairly similarly to the
64:41 prompt that we just did before for the uh pre-core report generator. And this
64:45 is again using another tool that I've created um for AI agent prompting. Um
64:49 so, it's fairly similar stuff that I I include in that other prompting tool,
64:53 but the agents is slightly different um to just regular LLM steps within
64:56 different tools and workflow automations. So, I will include this as
65:00 well. It's my AI agent perfect prompt generator and it's fairly
65:02 straightforward to use. I'll include that in there as well. But basically,
65:05 you put in all the information about what the agent is, why it's doing it,
65:07 the different tools that you're connecting to it, and then it prints out
65:10 this for you. So, I'll just run through it. We've got a role here telling it who
65:13 it is, um, and kind of hyping it up and saying how good it is. Explains the
65:17 task, um, talking about how it's helping to conduct detailed research on
65:20 companies and prospects. Um, some specifics. Uh, don't need to worry about
65:23 those too much. Just reiterating the task. And now here we can enter in the
65:27 references to tools. So if we go slash tool and see in order to get the agent
65:31 to function as well as possible, we need to tell it what tools it has available.
65:35 This is really key across all the agents you build, especially if they're more
65:38 conversational. You need to explain to them what tools they have and how and
65:41 when they should use them. So if I go, company. Um, yep, that's right. Purpose,
65:50 input, company URL, use when needing to gather company information. This one of
66:01 prospect. And then this last one is our [Music] pre. There we go. So that's all whacked
66:05 in there. Don't need to worry about that too much. But again, this will be
66:08 included um this prompt and everything if you want to follow along and also if
66:11 you just want to clone the whole agent um and use it in your own business or
66:14 sell it or whatever you want to do. We also get to select the model here. Um
66:18 I'm just going to keep it as GPT4 mini. I like some pretty fast responses here
66:21 because agents as someone's using it, it can feel really irritating if it's not
66:25 responding quickly. So, we've got all that built out. That's the core
66:29 instructions in the prompt of the agent. Um, we can go down to the uh tools
66:33 section. It's got all the tools connected in here because we mentioned
66:36 them in the prompt. And we can just go through and do some quick settings on
66:39 here. Um, I don't want to have to do an approval for it. Some tools you can say,
66:43 look, they've got to give it a thumbs up before it can actually uh trigger it.
66:46 Um, prompt for how to use. Just some quick descriptions we can pop in here.
66:49 I'm not sure why relevance doesn't carry that over from the tool. I guess they're
66:52 asking us to do it again for some reason. Um when you need to research a
66:55 uh prospect linked and URL, we're going to say this is auto run as well. Um and
67:05 then preall change this to auto run as well. Use this when you need to generate a pre
67:15 call report from the company and research. All righty. Um, and there's
67:21 all sorts of other cool stuff. Relevance, as you can see, is like
67:25 abilities, sub aents, metadata, extra stuff that you can build onto. Um, but I
67:27 just want to get you guys started with the core of this. Um, all right. So, we
67:36 that. And boom, we have our agent here ready to go. So this is where you can
67:38 test your agents and use them if you want to. But in this case, I'm just
67:42 going to give it a quick rundown and say see if the functionality is working as
67:47 we as we planned. Um, hi, I am getting on a call with Liam Otley from morning
67:55 side AI. Here uh has lenol report please to prep for the call. Sending your task to big boy sales
68:24 copilot. And then we get to see all the debug and how it's actually walking
68:28 through these different steps. Oh, let's does. Yep. Okay, that's great. It's
68:36 using the research company as we wanted it to. Should add in one more step there to
68:42 research the prospect as well. There we go. Using the second
68:46 tool. It is pretty satisfying when this stuff works. And this is just a really
68:49 basic one, guys. I don't want you to think this is like, oh, well, that's
68:52 pretty underwhelming, Liam. I'm trying to teach you the basics so that you can
68:56 actually build on top of this. So if you get the bug, if you get the like you get
69:00 a travel bug, if you get the agent bug and you see the stuff and you're really
69:03 interested. Oh, look, there it is. Now it's filling out the prospect summary as
69:06 the inputs. Surely we don't have to watch it do that. It's going to take a
69:11 while. [Music] Um, I really write that out word for word like that. But when you start to
69:18 see this magic and you add in other cool tools and functionality, you test it on
69:21 yourself. You can build like things for maybe you want to do content, you make
69:25 yourself a little content co-pilot, etc. There we go. To use all the tools
69:29 and it should be spitting back and bam, there we go. So, I hope that was worth
69:33 the wait. Let's go through it now. Here's your comprehensive precore report
69:35 for your upcoming conversation with Liam Mley from Morningside AI. Pre-core
69:39 report. Talent acquisition under pressure. Morningside AI operates in a
69:42 highly competitive AI and tech market as they scale. Finding specialized talent,
69:46 engineers, data scientists, AI with a proven record can be challenging. That's
69:50 scarily accurate because that is literally one of the biggest constraints
69:53 that we have had to scaling Morningside long-term is that it's just really
69:57 really hard even with my channel. It's so hard to find the right people and get
70:00 them to commit as as developers as well. So if you want to build a very big
70:03 general AI development firm, an AI automation agency, you need the best
70:06 talent and you need to get a lot of it in um so that you can scale up. So
70:10 that's bloody spot on. Obviously this thing knows that's a good angle to sell
70:15 through. Um but yeah, prospect analysis. Liam's a dynamic entrepreneur and
70:17 thought leader with a robust background in e-commerce, digital marketing and AI.
70:22 His journey reflects a passion for innovation and commitment to continuous
70:26 learning. Again, that's pretty bloody spot on um hands-on experience. So there
70:31 you go. That is the big boy sales co-pilot for big boy recruits. The cool
70:34 thing you can do now once you have built this um is you can go share um there's a
70:40 chat UI which I'm going to turn on now. There are chat widgets so you can put
70:43 them on websites and stuff like that. What I want to do is just pull this up
70:46 because this is what you'd be giving to your client likely or if you guys are
70:49 going to start selling these to businesses which again we're touching on
70:52 selling in the last section of this video. So how do you turn these into
70:55 into a business and start making money from it and selling these as a as a
70:58 service and building these businesses which is really where the money's money
71:01 is made. So there you go. This URL you can obviously send to your client. If
71:03 you're building it for your own team, you can send this to your team and say,
71:06 "Hey, pin this because you're going to be able to use it. Add more
71:09 functionality into it, etc. That is how All righty, that is build number one out
1:11:12 Build 2
71:20 of the way and we are jumping into AI agent build number two, which if you're
71:23 listening closely at the start of the section, we are talking about an
71:28 NATbased inbound lead qualification agent that's going to be doing some
71:31 pretty cool stuff for us, which is a really important function within a
71:34 business um around lead qualification. Um, so this is a really cool one. Again,
71:37 in this case, this is what we call an automated agent, not a conversational
71:40 one. What we just built is a conversational agent. humans are
71:43 directly talking to it and chatting back and forth and using it and we are
71:46 operating it ourselves. In this case, as you can see on this little flow uh flow
71:49 diagram here, this is a screenshot from the final product. We are actually
71:53 baking this into a workflow that's going to be triggered on a form submission.
71:56 We're going to do some research and then we're going to use the handy AI agent
72:00 and tools agent within NA10 to trigger another workflow and then send some
72:04 emails off. So, this capability of using AI agents in workflow automation really
72:08 expands the possibilities of what you can build. And the software NAN that I'm
72:11 going to teach you how to use is really at the the cutting edge and leading the
72:14 charge when it comes to these automated uh AI agent workflows. And just quickly,
72:17 it's good that we walk through the purpose and the value behind this
72:19 automation so that we know why we're doing it. Right? So this inbound lead
72:23 qualification use case is based on the fact that companies who market
72:25 themselves well soon have far too many people reaching out to them. Many of
72:28 which are not a good fit or what you'd call qualified for what they sell. Eg
72:32 they're too small or they're not the right industry. Like you you have a
72:36 business and they say we only help XYZ kinds of businesses. And if leads come
72:38 to that business who are not qualified, then they obviously don't want to be
72:41 taking calls or or doing anything further with them. So this process of
72:44 researching a new lead and deciding whether or not to take a call is known
72:48 as qualification, which is what this agent aims to automate. So the value
72:51 here is that instead of having to pay someone to manually qualify and go
72:54 through all of these leads or using arbitrary rules, which is what some
72:57 businesses have to go to, it's like look, oh look, we've got so many leads.
73:00 Let's just say if they don't want if they say that they're not on this, then
73:03 we'll just cut them out. And that's potentially leaving money on the table
73:05 by cutting out leads who would have actually been a good deal, but the rules
73:09 kind of didn't see enough detail to be able to determine if they're a good fit
73:12 or not. So, this automation is essentially immediately qualifying and
73:15 triggering the next steps to the sales team and allowing them to do that human
73:19 style research on these leads at scale. So, enough talking. Here's a little bit
73:22 more information. Again, with all of these, it's going to be on the figure on
73:24 the school. Then, I've also broken down how this agent would actually operate in
73:28 the real world and sort of real time. A lead's going to be submitting the form,
73:32 which is going to be this here. um the relevance company AI researcher. So
73:34 something that we've just built in relevance, we're going to reuse in here,
73:37 which is handy that you can start to move these components around and see how
73:39 they can fit into different automation platforms. Um then the AI agent is going
73:43 to look at this information from the research. Then it's going to determine
73:46 based on a qualification criteria we give them um inside the prompt of this
73:51 agent whether they are qualified or not. If they are qualified, it's going to uh
73:55 use this tool here and call this second workflow. when we're going to
73:58 essentially analyze that further and do a notification to either our agency team
74:03 or our our SAS team and if they are not qualified we will use this tool here
74:06 which will just send an email straight back to the person who submitted the
74:09 form say hey sorry we're not open to working with businesses like yourself at
74:12 the moment let us know if we can help you any other way so that's a rough
74:15 rundown of the build let's jump into it so to kick things off of course you need
74:20 a platform on natn.io io. All links and resources will of course be on the school uh post for this
74:24 video. And once you're on this page, you can go to get started and you can create
74:29 an account for free and just go through the setup process that they do. I'm sure
74:32 you can figure that out. They do have a 14-day free trial as of this filming. So
74:35 that's very good if you're just jumping in, not having to pay anything. And they
74:37 give you quite a lot of usage up here. As you can see, 1,000 executions. So
74:40 what we're going to do, of course, is click on create workflow up here. I will
74:43 be giving you the template. So if you want to just import it, you can. That
74:46 will be on the Figma there. But just like the last tutorial, I'm actually
74:49 going to be showing you the process of building these up from scratch that you
74:52 can see how I how you go through the process of building these automations
74:56 and the the testing and back and forth you need to do in order to get to the
74:58 end result. That's probably actually a lot more important if you are to go out
75:01 if I'm trying to teach you to fish, not just give you a fish, is to see how I
75:04 deal with problems when we're building these. So, let's get started by starting
75:09 off our trigger. We're going to go form has a nice form on new inn form event. And here we
75:17 get to create an N8N form. So we can call it uh work with your. So now we get to pick the field
75:31 names. So we want to have the first one is make that a required field. We add
75:40 another one. What is your company website? field. Right. So, we've just built out a
75:55 basic form here with first name, company website, which we're going to need for
75:58 the next step. I've put a placeholder in here so they know that it needs to have
76:02 https um at the front of it. And I'm asking for them to provide some
76:05 information about your inquiry like and maybe you can say what can we help
76:10 you? Um and that's going to be text area. So, they've got a bit more room.
76:12 So, we can go test step here. Make sure that it's all looking nice. This is what
76:15 the form's going to look like. What's Obby. There we go. We've submitted that
76:24 form. If you go back, there we go. We have the data in. So, this shows you the
76:28 output. NAM works by having kind of this middle island here, which is what you
76:31 set up. And then the left side is the input and the right side is the output.
76:34 So, we've tested it and these are the outputs that our form is giving, which
76:38 is what we are looking for. And the next step in this lead qualification process
76:42 is to do some research on the lead. So we have the company URL and this is when
76:46 we're going to go and make an HTTP request. So this is basically calling
76:50 any API over the internet. And in order to set this up, we actually need to go
76:57 relevance and we're going to find the company researcher tool that we made
76:59 with relevance. And now you're going to start to see how this all fits together.
77:03 um that building tools and relevance can also be very useful and an extremely
77:07 useful skill in all areas of AI automation because now I can come on to
77:11 research company and not only can I use it here this I mean this is why I think
77:15 relevance is such a great platform um you have the use here so I can send this
77:18 across to a client I can share it with it as I showed before I can use it just
77:22 here myself I can run it in bulk on a spreadsheet or but more importantly in
77:27 this case I can go to the API and now I can call this might look scary just
77:31 don't worry I can call this functionality basically send in a
77:35 company URL and get back the research. I can access this over the internet
77:38 through an API and they give me it here and they tell me exactly how to call it.
77:41 So now we this is on actually a post request. So we can copy this. Remember
77:46 how we talked about get and post request. This is a post request because
77:49 we're posting some data to relevant AI. So we change this method to post. We put
77:54 the URL in here. We do not need authentication in this case but you can
77:57 turn it on. So you can make a private here and then you can have an
77:59 authenticated. might sound a bit complicated but for now don't worry
78:02 about it we don't need to have an authentication step on this and then if we look at the request
78:09 body here it tells us how we can send data to relevant AI and if we go copy
78:16 here come back we're going to send a JSON and we can change this to using
78:26 JSON and then we can paste in basically what we have been given from relevant AI
78:29 now this looks a messy. Let's pop this open a bit more. And we actually need to change it
78:35 to an expression here. So fixed means that we're not accepting any dynamic
78:39 data in. So when the form is submitted, we actually need to take in some data
78:42 from that form which we have over here. We need to inject it into this uh HTTP
78:47 request to relevant AI to get this company research done. So we can't have
78:51 it as a fixed um JSON body here. Need to change it to expression and then we can
78:54 pop this out here and it gets a bit easier. So, we have params. We have the
78:59 company URL. And then we need to pop in here. Oh, the company
79:05 URL. Pop that in there. And here on the right side, we get to see what that
79:08 would look like given the test data that we've just put through. So, you can see
79:12 I've got the company URL, https morningside.ai in quotations here, which
79:15 is what we want. So, we can go back now. And now we can give it a test to see if
79:18 it's going to be able to communicate with relevant AI and get us the data we
79:24 want. There we go. We have our result back from relevant AI which is the
79:28 summary. As you can see when we go back to this um and back to build this is
79:33 exactly what we'd expect. You know you put in the URL does this scrape a fire
79:37 call writes the summary spits out the summary and this summary is what we're
79:40 getting back out over here which is what we want. So bank that's great another
79:43 step done. And those of you who are a bit confused about what this is this is just
79:48 the body of the request. So because we are sending a post request remember how
79:51 we have get and post request. Get requests are typically just with a URL
79:54 and with a bunch of stuff tacked on. A post request, we need to send a a JSON
79:59 body like this. And it does look quite confusing, but if I take this and go
80:09 form, paste this in and beautify it. And you can see how we have basically the
80:12 project which is the relevant project that we are calling. And this tells the
80:15 API this is the project that we want to interact with. and the project expects
80:20 the params, the inputs, which is the company URL, and we're injecting that
80:25 company URL from our information here in NA10 that we've dragged across. It might
80:28 seem tricky a few times, but trust me, this stuff becomes like riding a bike
80:31 once you get up and running. So, um, a few more of these and you'll be you'll
80:34 be completely fine. All right. And so now that we have our company
80:36 information, the next step is going to be setting up the agent, which is really
80:39 the coolest part in my opinion about make right now is that we come in here
80:44 and we can click on um agent and we can set it to as a tools agent, which means
80:48 we connect our own tools. And if we just back out of this, we can do cool things like set up
80:53 the model. And right here, this is so cool because we get to see exactly what
80:57 we were just talking about earlier in this video, but we have the different
81:00 parts of an agent, the different ingredients, right? So here's the chat
81:03 model. This is the brain. This is the LLM that's going to be powering the
81:06 whole agent using the soup example. Like this is the one meat that we get to
81:10 choose, right? This is a specific model. In this case, we going to be using
81:13 OpenAI again because you already have your API key and I can't really be
81:15 bothered going and showing you a whole another provider, but it's the same
81:18 process for all of them. You can if you want anthropic, you can then pick all
81:22 the anthropic models. Uh but in this case, just to keep it simple, let's just
81:30 to and open AI model. And if we go back again, you can see that we now have uh the memory and the
81:35 tools that we can connect. So, of course, we have the tools, which we've
81:38 talked about a lot in this video already. We can connect multiple
81:40 different tools here, as we're going to do in a second. And then we also have
81:43 memory set up here, which is a little bit outside of the scope of this video.
81:46 As I said, in most platforms, it comes builtin, but NAT is a bit more of a
81:49 developercentric platform. So, if you wanted to play around with memory, um
81:52 different forms of of managing memory, you can do that here. In our case, we're
81:55 not going to be touching it. And in order to connect a knowledge base, if we
81:58 wanted to set up a knowledge base for our agent, we would connect it as a
82:01 tool. So you can see here we've got in-memory vector store, pine cone vector
82:05 stores, etc. These are vector databases that we can connect just like in the
82:08 other tutorials we're going to do in this video. When you upload some
82:10 documents to make a knowledge base, they're essentially being put in a
82:13 vector store like these, but the platforms manage it for you and make it
82:16 a lot easier to do. So in this case, we're just going to be doing two
82:19 different tools. Firstly, we're going to be calling an NATM workflow. So I'm just
82:23 going to finish off the the basic setup of this. Um, and then we're going to add
82:26 another tool on. It's going to be the Gmail tool. And then before you know it, we
82:33 have our AI agent structure built out. So, we're using the Open AI models.
82:36 We're going to pick the model shortly. We're going to be using the tool to call
82:39 the second NATM workflow, which is going to be uh triggering the the email
82:43 notifications for our sales reps and the classification of the uh of the lead.
82:46 That's going to make a bit more sense when we actually do it. So, just stick
82:49 with me on this. And then this Gmail is going to be sending back a hey sorry you
82:55 didn't qualify for what we do um sorry we can't help you let us know if we can
82:58 do anything else for so to start setting things up we can start from left to
83:01 right here with the chat model um the openai model that we want to use and you
83:04 need to set up your openai account so you can click create new credential and
83:07 you need to go and add in your API key here I'd suggest you go and make a new
83:11 one on platform openai.com and you can add a new one in here and you can name
83:14 the key nat so you start to know which keys are used for which different
83:18 platforms um and once you put that in there you can Just click save and then
83:20 it will run a little test and then you're ready to go. You should have this
83:23 set up here. Then we get to select the model that we want to use for our agent
83:26 here. And in this case, I want to go for something quite smart. So, I'm going to
83:28 go for 03. We have 03 mini here. This one appears to be a little bit more recent.
83:34 So, they'll sometimes put the dates on the end of it. And if you just look at
83:36 the current date, you can kind of see how how close to the current date it is.
83:39 Um, but I'd say this one's a bit more recent, so it's going to be hopefully a
83:42 bit better. And next, we're going to skip this cuz we need to set up a whole
83:45 different workflow to connect it to. Um, we're going to jump straight into this
83:47 Gmail one. You need to set up again another connection as you go through all
83:50 of these different automation platforms. You do need to do these these
83:54 connections between uh your own account, say your Gmail or your calendar or all
83:57 these different apps. You need to go and create a new credential. Um and you can
84:01 just do the sign in with Google here. Super easy to do. I'm sure you don't
84:04 need my help with that. Once you've set up that connection, you can close this
84:07 and you will see the connection that you set up here. Basically, that's what
84:09 we're going to be sending emails through. And then we can get into
84:12 setting this up. So, because this Gmail tool is going to be used to send a a
84:15 reply back to the person who submitted the form and say, "Hey, sorry, you're
84:18 not a good fit for us." You want to send it back to the person who submitted the
84:21 form. Now, if you scroll down here, uh, yep, you see that I've I've forgotten to
84:25 add the email form in. So, this is a good example of needing to go back a
84:28 little bit. So, we can go back to the form submission here. Um, scroll down,
84:35 say, what is your email? And we can set it as an email. So, it's going to automatically force
84:41 them to provide a valid email for us. And I want to maybe shimmy this up a
84:49 Um, so it's name, email, then company, website. Um, we can do another test here
84:55 just to give it some proper data. Oh, it's not going to let us do
84:58 that because this isn't set up properly. So, we can just delete that for
85:03 now. And we're actually going to delete that. Otherwise, it's going to be a bit
85:06 of a pain. again. So, we can submit another form here and go back to NA10. And there we
85:20 go. We have the information. We have the email now. That's great. And so, now we
85:23 can come and set up our tools again. We've got the workflow there and we've
85:28 got the Gmail. Um, and now we have our Oh, and we haven't got the data here
85:41 because we need to run this again and get the research. So, now the
85:44 research is done in there for us to set up the Gmail tool here. We can go to two
85:49 and we'll be able to pull in the email. So, again, they submit this form. We
85:52 realize that they're not a qualified uh person for our offer. And then we're
85:55 going to send an email back and say, "Hey, sorry, you're not a good fit." So,
85:59 we can say the subject here is um thanks for your interest. I'm going to change
86:04 the email type to text here and I'm going to write a basic message in. I'm
86:07 just going to snag it from the one that I've done previously. So, we need to
86:10 change this from fix to an expression because we want to be pulling in their
86:13 name here. So, I'm going to paste in what I have here. Again, this would be
86:16 included um in the resources. It's just going to save us time if I don't have to
86:19 type this all out manually. Um but you can see here, I'll just delete this so
86:23 you guys know what we're doing. If I go hi, or hi, and then we can add in what
86:29 is your first name? Hi, name. In this case, it's going to be filling in my
86:32 name here as an example. So, highly mly, thank you for your interest in big boy
86:35 recruitment services as you specialize in recruitment for software and
86:38 development agencies. We're not a good fit based on your company's industry.
86:40 Please let us know if you'd like to connect us with one of your partners who
86:43 specializes in dealing with your needs. Cheers. Huge Jackman here to sales, big
86:48 boy recruits, BBR, uh, Dallas, Texas. So, if you guys remember huge Jackman,
86:52 comment down below um for the OG fans. And then that is our Gmail all set up.
86:56 And just quickly so that you've got a bit more knowledge around how this Gmail
86:59 uh tool works, we have all these different steps that we can use. We're
87:02 using the send one. So it's sending an email. You can use reply, you can use
87:05 get, delete, all these other functions, but the easiest one and the most common
87:08 one you're going to use is going to be that send one, of course. Now, we need
87:11 to set up this NATM workflow which the agent is going to call as a tool um when
87:15 they are a qualified prospect. So I'm going to delete this one and just save
87:18 this for now. Then we're going to go back to home. We're going to create a new
87:25 workflow. And this is a really cool skill that I want to teach you. The fact that you can
87:29 build all these workflows and then connect them to agents and it can just
87:32 be taking data in and kind of shooting it off in all directions and triggering
87:36 all these complex multi-step processes because it's a super valuable way of
87:39 using agents. Um, so I really want to teach you that. And obviously this one's
87:42 going to be starting off a bit different to the other one. We actually want this
87:45 to be set up as when executed by another workflow. So that's going to be what the
87:49 trigger is here. And we are going to be able to define using fields below. Let's
87:54 just add in one here that is a lead lead name. Um, for now we can just leave that
87:59 there. But that's all that we need to set up. We need to go back over to the
88:01 other one. Just needed to set this part up. Let's rename this qualified lead
88:07 lead classifier and notifier. So we can save that there. And
88:10 we have a bit more work to do on this other one. If we go back to this, we can
88:16 rename this here. So let's call this our lead qualification agent.
88:23 And so now we have our other workflow set up. We can come here. We can call
88:27 another workflow with the tool and we can call it lead is qualified. And so
88:31 now is when we get to tie back into what we learned in the foundation section
88:34 because we are now writing descriptions for our tools. Remember how we had
88:38 schemas and scheas are basically written instructions or instruction
88:42 manuals on how to use tools and how to use the APIs that wrap around them. Um
88:45 this description is going to be basically those descriptions that you
88:48 put in the schema. But NAT is going to be basically constructing it for us in
88:52 the back end. And we just get to put in here, okay, what's the name of this
88:55 tool? It's going to be called lead is qualified. It's giving us a nice example
88:58 of how we can write a description for the tool. So call this tool to get a
89:00 random color. The input should be a string with a comma separated names of
89:03 colors to execute. So in our case, we can say call this tool when the lead is
89:07 qualified according to our criteria. The inputs should be lead name, lead email,
89:18 company, company summary, and request in info. We're basically telling that AI
89:22 model or the brain what this tool can do. So when we send it some kind of
89:25 input, it then it looks over our tools. It looks at the Gmail description and it
89:29 looks at this uh workflow tool description and goes, "Oh, well, I have
89:31 a tool that does this and a tool that does this. What have they just sent me?
89:34 Okay, now I think I know what I need to do from here." So this is the rough gist
89:37 of what we want to do as a description for this tool. I'm actually going to
89:40 beef it up with a bit of a a bigger one here. Um if the lead is qualified to
89:44 work with big boy recruits, eg they are software based business like SAS or
89:47 development agencies and trigger this tool and send the lead data in the
89:50 following format. It's just dummy data. So name a name email um an email message
89:57 I want new div qualified true company information and company information
90:01 which is a summary of the relevance tool to do the company research that we have.
90:04 So, might seem a bit crazy at the moment, but stick with me because it
90:07 will make sense in a second. We're basically just told it when it's going
90:09 to trigger this tool and the format to send the data in. And then for the
90:13 workflow, we get to choose here the one that we just set up, which is our
90:17 qualified lead classifier and notifier, at least in my case. And then we see the
90:20 workflow inputs that we've just set up. So, if we go back over to our other
90:24 automation, so when we open this up and define our inputs here, you can see over
90:28 here we are getting just one of them that we've put in as an example so far.
90:32 So now we need to set up all these inputs correctly. And we have the name
90:35 that we want. So we've got the name and message. Um, honestly don't think we need this
90:54 qualified one here. And then we have the So, if we just test this, head back over
91:13 we refresh this list. Oh, back out. Save it. And this pops up. It says that these
91:21 inputs are outdated. So, there we go. We have lead name, email, message, etc. And
91:24 then we can actually automatically fill out a lot of these inputs. Maybe I will
91:27 put this back in here just to show you qualified. Um and then we just call it
91:39 uh true And we go back here and we add in one more which is a
91:58 qualified which is also a string. And so we have this qualified field here as
92:02 well. If we go back um I'll just test this. Save it again. Come back over and update
92:09 this. I think we can actually make it even cooler. So let's go. Um it changes
92:13 from a string to a boolean. So that's either true or false. Um, and if we test
92:20 this, save it again, and we change this to take away these little things. Sorry,
92:24 pisses me off if I don't have this set up right. Um, and then we update this,
92:28 you'll see this turns into a switch. So, that's a true or false, right? In order
92:31 to trigger this tool, it should be qualified by default. So, it's a little
92:34 bit redundant, but it's still cool to show you how we can get AI to fill out
92:37 these fields. Um, in a lot of cases, we don't, but for this qualified run, we
92:41 can. So, we can set this as an AI generated field. We can say if the elite
92:47 is qualified based on our criteria this set to true. And then for the rest of
93:07 quickly. Give this a second. Open this back up. And we can fill out some of these
93:14 fields. So we can go lead name that we want to pass through to the other
93:17 automation is going to be that. So we can fill a lot of these in email
93:26 um message and the company information um we can get from um this technically
93:30 but um maybe we could do a cool AI generated one here which is um a short
93:37 summary of the company and the industry that they are in. company's details. So, this is basically
93:46 telling the AI model of our agent how to fill this field out, which is one of the
93:49 reasons that AI agents are so powerful. So, that's all set up. Now, we've got
93:52 all of our tools set up. The last step is just to set up a prompt for our
93:56 agent. And I am going to cheat a little bit here and just throw one in that I've
93:59 done before to save us a bit of time here. So, we want to set the prompt
94:05 here. We can paste in this information here. And this is basically just telling
94:08 the agent who it is and what it's supposed to do. You're a lead
94:11 qualification agent. Your job is to analyze the form submission and company
94:14 research provided and then decide whether they are qualified to work with
94:18 big boy recruits. Ra we specialize in XYZ. Um we are specialist in capturing
94:22 talent for ra. We only work with softwarebased businesses, EG SAS
94:25 companies or development agencies. These companies are willing to pay much more
94:29 developers than your average marketing company or local business. Therefore, we
94:31 only work with them. Your job is to determine if the lead you are provided
94:35 with is a good fit for big boy recruits. And if so, call the lead is qualified
94:38 tool and send the elite information to it. If lead is not qualified, then you
94:41 must trigger the Gmail send email tool for us to respond to them letting know
94:44 letting them know we are unable to work with them. And then we have a response
94:47 format here which we can probably just delete. And then we can add in here is
94:50 the lead to information for you to analyze. Let's pop this out to make it a
95:07 We can add in um just go name um company URL company and we take it. So that's from
95:22 the relevant step for the research that we did to scrape using our firecrawl
95:25 tool. And then we provided all the information to this agent and it's going
95:29 to be injected with all of these values on each form submission and then it's
95:32 going to make a call on what tool it needs to use. So we're pretty much
95:37 there. We can even give it a run here and try to test the step and see which
95:46 choose. If we go back we can see okay look it's used the chat model as the
95:51 brain and it's triggered the NA10 workflow as expected. You can see here
95:54 that it's sent off information to our other workflow. It sent the lead name,
95:59 the email, the message, and the company information, and it set it as qualified
96:02 as well. So, all of these fields have been filled out. We've got a nice AI
96:05 generated summary here from the model and brain. And we have the qualified set
96:09 to true. And so, the final step now for us is to head back over to our other
96:14 automation and just finish it off. Oh, we need to save that. I will
96:19 just run that again for you so you can see it in slow motion. It's using the
96:22 LLM as the brain and the tools agent and it's deciding whether it's qualified or
96:26 not. And if it is qualified, then it will send it to this workflow. Bam,
96:31 we've sent it. And there you go. If we this. Head back over to our qualified
96:40 lead classifier notifier. Now, we can add on a quick few steps here. I'm just
96:43 sort of going to rip through this. Um it's not super important. Um but it just
96:46 shows you a little bit more functionality of what you can build in
96:49 on N10. So, we're going to add in a messenger model step here, and we're
96:57 mini. And what I want this to do is to take in that information that we sent to
97:00 the workflow about the company research, etc. So, we know this is a qualified
97:03 lead now, but we just want to split it between either our SAS team or our
97:07 development agency sales team. So, they're specialized in dealing with
97:09 different cases. So, I'm going to cheat and just throw in a prompt here, which
97:13 you guys will be able to get access to. Um, which is basically saying we have a
97:17 new inbound lead. Um, change this to an expression. Sorry. Um, we have a new
97:20 inbound lead that we need you to classify into either SAS or development
97:24 agency. Here's the lead information. Um, we need to go back step and test
97:30 this. There we go. We should have some information. Um, and now we can put
97:34 these in. So, you see how there was nothing here before I went back and
97:36 tested the trigger so that it gives us some null values here that we can fill
97:41 out. Here's lead information. Um lead name uh message um name
97:51 request company information if the company is a SAS output SAS if the lead has development
97:57 agency upput agency. So we're looking for just agency or SAS as the outputs
98:04 here. Um simplify the output. Yep, that's all good. So we can no point in
98:07 us testing that step there because all the values are null. Um but the next
98:12 step is a basic router flow if so this is a basic conditional
98:19 routing. So we have the conditions we can go expression here. So we can go
98:24 um the content here. So this is the output from the open AI step. If the
98:29 content which is the response from the LLM step the classifier it's either
98:32 going to be agency or it's going to be SAS. So if it let's just to to make it a
98:36 bit more flexible. If we go string, if agency, great. So, if it contains agency
98:46 on the true side, we want to go Gmail and we want to send a message. Um,
98:54 and then if it is false, we want to do basically the same thing. Now, I've got
99:11 Um, okay. So, here we're not getting much data on the input side here and we
99:14 can't seem to simulate it because it's of course triggered by another workflow.
99:18 What we can do is just save it here. go into executions. And if we go back to our
99:29 agent, and we can go to the form submissions one. If we go to executions, and we just
99:36 run one of these that we just did before, copy to again. This is basically just going to
99:48 trigger this again. and so that we get a fresh execution and we can sort of pull
99:52 that data back into the workflow. Oh, again. Boom. Triggered it. And that's
100:28 all done. Now if we go back to this and we go to executions, go to the most
100:32 recent one that succeeded. It's going to load in. Oh, hang on. This one's
100:51 it. Okay, so this one here, if we click this, yep, we've got all the data in
100:55 here. So, what we can do is copy this into the editor and then we've got the
100:58 data that we need that's already loaded in so that we have some values to put
101:02 into our Gmail steps Gmail. So, that's a handy little trick to to know how to do.
101:07 And now we have all of this information. So, that's what that's what I was trying
101:11 to get. Um, the same setup and we're going to send this to I'm just going to
101:15 use an example here and call this um it's the same email. You wouldn't pull
101:18 this in necessarily. I'm just using this as something that I can show you
101:24 at show. Say new agency lead. Let's do a text. We say um new agency lead man. Go
101:30 get him. Um turn into an expression. And then we say we can just
101:37 throw this company data in there. It's going to be messy. You can play around
101:40 with this more when it comes to formatting, but just to show you the
101:44 functionality. If we go uh test a step here, that's going to sent an email to
101:49 this. This is like my agency sales reps uh email. Of this. Can duplicate this. Right click,
102:06 duplicate, bring it here. Oh, connect this up. And we change this.
102:14 You change this to your like SAS guy. You change it to a different different
102:18 email. Um, of course, and then you can say new SAS lead. Right. So now we have done all of
102:27 that basically all built out. The data is going to come in from the agent. It's
102:30 going to send in the company summary. This is going to classify it into being
102:34 a uh agency lead or a SAS lead because those are the only two types of
102:37 businesses that we work with. So all of them will be qualified when they come
102:40 through here. And then it just sends an email to our uh agency sales rep or our
102:46 uh SAS oh rename this to our SAS sales rep, new SAS lead for them to continue
102:56 with. Right. So to test this we can turn this on to active and you can see that
102:59 you can now make calls from your production form URL. Um we can go okay.
103:05 If we double click on this we can open this up. We can click on production URL.
103:11 Copy this and open this up in a new tab. spin. So of course my agency Morning
103:20 Side AI does development services. So this should be qualified and it should
103:24 also route it to the agency email. So if I now go submit, we go back into NADN, we go into
103:32 executions, we can see this one is running. If we go to inbox and there we go. We see it has
103:47 succeeded. And then if we go to and then if we go to our lead
103:51 qualifier and notifier and we go to executions, we will also see that we
103:55 have a new one that has succeeded here which was just uh a few seconds ago and
104:01 that's gone through. It is um outputed it as agency which is the the
104:05 classification that we wanted. It has gone through and has sent a new email.
104:08 And if we go to here we have new agency lead there. There we go. All the
104:11 information. So that's working number one. Now we can go back to our form and
104:14 we can try it again but this time with let's say an unqualified business. Let's
104:22 go. What is your name? Ray Croc Ray McTum I need more guys more people
104:33 flipping damn burgers. So essentially Ray here has come to our recruitment agency and
104:39 they're asking, "Hey, I need people to do flip patties for me in my fast food
104:43 restaurant." Um, and because Big Boy Recruits in Dallas, which is a
104:47 hypothetical company, of course, um, doesn't do that. It's going to qualify
104:50 them or it's going to disqualify them and then send an email to our good
104:54 buddy. Oh no, some poor dude at McDonald's is going to get an email now. Um, because
105:00 send an email. Um, but it's going to be running and of course it's going to be
105:04 sending. if they're unqualified, it will send an email to them and say, "Hey,
105:07 sorry you're not a good fit for us. Let us know if we can help or we can connect
105:13 partners." And while that is running, I would just put together the final one
105:16 here to test the functionality, which is if we go Liam admin.com, and we set up my SAS
105:24 https, my SAS agent, if you haven't already used it, we're going to show you
105:26 how to use it in the last tutorial of this video. So, you guys will get to see
105:30 that. Um, which is my own no code AI agent building platform. And what can we
105:34 help you with agents? Um, so this should be a SAS one and it's going to qualify
105:45 have the McDonald's one has succeeded here. And you see, yep, as expected, we
105:49 were not qualified. The McDonald's person was not qualified for our offer.
105:52 So, it looked at the qualification criteria we provided in here, said,
105:55 "Hey, no, that's not a good fit." So, I'm going to use this tool. And you can
106:00 see that it sent the email and it said, "Hey, thanks for your interest. Um, but
106:04 we're not a good fit for you." So, someone at McDonald's just got an email.
106:07 Apologize for that, but we didn't trigger the other workflow, which is a
106:10 key part. And we're not going to send emails to our sales team saying, "Hey,
106:14 look, new leads." Now, I have sent another one through here which just
106:17 finished executing. And we can see this. It's gone through. It's researched um
106:24 agentive. you'll see um Agent is a leading service delivery platform for AI
106:29 agent AI automation agency owners um etc and it's called the tool because we were
106:32 qualified because we're a SAS business right and again if we go back to
106:38 here and we look at the most recent one here then you'll see new SAS lead has
106:48 been triggered because we are a SAS of course um the LLM step here has outputed
106:53 just SAS So that means that it should send an email to the SAS team, which if
107:02 inbox, tada, new SAS lead, right? So I know that may have taken a while,
107:06 but uh we got there eventually. And you can see that we've built out all of this
107:09 functionality. We have our AI agent calling our tools if they are qualified
107:12 and triggering this other workflow. Again, you can build so much cool stuff
107:15 by connecting an agent to multiple different workflows. We have a little
107:18 relevance AI researcher tool that we're reusing here and we have people getting
107:22 denied um with an instant email sending them back. So, hope this been a cool one
107:26 to show you how NATM works. I really, really like this agent functionality
107:29 that they have. I think you guys are going to be able to build some awesome
107:31 stuff if you keep going down this rabbit hole. So, that has been agent build
107:35 number two. Stick with me as we jump into agent build number three, which is
107:39 a pretty damn cool one, focusing on both chat and voice-based agents all in the
107:43 same build. So, let's get the ball All right, so that is two builds out of
23:00 Advanced Tools Use
23:00 given multiple tools to work with. Now, obviously having an AI agent
23:04 that just capitalizes text isn't very useful. I get that. The real magic
23:08 happens when you give agents multiple tools and the ability to use them
23:11 together in order to achieve complex goals. So, do you remember our
23:15 definition? AI agents are workers that can understand instructions and take
23:19 actions to complete tasks. When you give an AI agent a task, it's going to try
23:23 its best to execute on it, but if it doesn't have the right tools on hand to
23:26 do the job, it's going to be useless. And so, the more tools that you can give
23:30 an agent, the more flexibility it has to solve problems just like a human would.
23:33 So, let me give you a real example from my own business, right? Say I build an
23:37 agent and give it the task. Find AI startups that have recently raised money
23:41 and put them in a spreadsheet and add a summary of each of the businesses in the
23:44 spreadsheet and then email me the link to the spreadsheet. When you give an AI
23:47 agent a task like this and provide it with multiple tools to use, it can break
23:51 down this problem just like a human would. For example, it might think first
23:56 I need to search for AI startups using my web searching tool. Okay, let's do
23:59 that first. Then I'll need to create a new spreadsheet with my Google Sheets
24:03 tool. And then for each company that I find, I'll need to add a row to the
24:06 spreadsheet. And then I'll need to write a summary of each business and put it in
24:09 a new column. And then finally, I'll use my email tools in order to send the link
24:12 to Liam. And that's all great, but then when you add on top of that powerful
24:16 reasoning models like OpenAI's 01 and 03 and even things like deepseat as the
24:20 brain of the agent that can plan, take actions, then reflect and then plan
24:24 again and so on. You have essentially created a truly intelligent AI that
24:28 solves problems and approaches them just like a human would. So, say for example,
24:32 the original plan was to use the web search tool to search for AI startups
24:35 raising money. Probably a terrible search term, but what if that doesn't
24:39 return any good results to the agent? Well, a human would go, damn, I need to
24:43 change my search term or maybe I need to try find a different method of finding
24:46 these companies on like LinkedIn or something. The latest in AI technology
24:49 like these reasoning models, it allows these agents to do this exact same kind
24:54 of reflection and replanning in order to achieve their objective. And this is
24:57 when you can really see why we call them digital workers because they can do
25:01 things like planning multiple steps. They will use different tools in a
25:05 sequence and even adjust their approach based on the results from those tools.
25:08 Now, I should mention that this technology isn't perfect yet, right? So,
25:12 these multi-step tasks are often unreliable and agents typically need
25:16 human supervision for more complex workflows. But things are moving
25:21 incredibly, incredibly fast. In fact, we're already seeing the next evolution,
25:24 which is multiple agents working together. Instead of just one agent
25:27 trying to do everything, you can have one main agent that you give orders to,
25:30 and then it can use all of the other agents underneath it as tools where it
25:34 can send specific instructions. Like underneath the main agent might be a
25:38 research agent, which is best at finding companies and has its own tools. Then
25:41 you have a writing agent that's really good at writing summaries. Then you have
25:44 an emailing agent, which has got all the emailing tools. And so each of these
25:47 agents can be specialized in their specific task with multiple tools. and
25:50 then they all work together to achieve a common goal. This is exactly what major
25:55 companies like HubSpot and Microsoft and Google are building towards. It's these
25:59 entire workforces of AI agents that can handle complex business processes
26:02 automatically. In the next chapter, I'll show you how to build AI agents like
26:05 this for yourself using no code tools. But first, we need to understand the
26:08 different ways that these agents can actually be used in the real
26:11 Conversational or Automated Agents
26:14 world. So, we understand how AI agents and tools work under the hood. Now,
26:18 great. If you don't, please go back and take some notes, right? You should by
26:21 now have a whole bunch of notes um from the stuff that we've covered already.
26:23 And this stuff that you're learning took me two years in order to learn and and
26:27 be able to apply effectively. So, you best believe it that it's going to take
26:30 you two to three watches before it all sinks in. So, if you're feeling a bit
26:33 lost and and overwhelmed, don't worry. That's how it feels with learning
26:36 anything new or how it should feel if you're learning something that's
26:38 actually pushing your boundaries and adding something to your to your
26:41 capabilities. Next, we need to look at the different ways that AI agents can be
26:45 used in the real world. There are two main categories of AI agents.
26:49 Conversational agents and automated agents. Conversational agents are ones
26:52 that humans interact with directly through chat on things like websites.
26:56 You've got maybe you're chatting to it on WhatsApp. You've got interacting with
27:00 it over the phone via phone call. You've got chatting to it via Instagram DMs or
27:04 custom apps and websites. For example, OpenAI's GPT platforms allows you to
27:08 create agents that you can chat with directly on your computer or on your
27:11 phone. or using platforms like my own Agent, you can connect these agents that
27:15 you build onto a WhatsApp number or onto Instagram. And I'll show you how to do
27:19 this in the tutorial chapter of this video. So, in all these cases, you or
27:23 someone else is there sending messages or instructions to the agent and
27:26 explaining what you want to do and kind of chatting back and forth with it,
27:29 whether it's on a website, WhatsApp, Instagram, or whatever. And within these
27:32 conversational agents, it's not just text based. It's like I said, there's AI
27:35 voice agents as well, which are an extremely exciting sector of the AI
27:39 space right now. And these systems use multimodal models that can take in audio
27:44 as input and then produce audio as an output. And so these agents can be
27:48 chatted to over the phone or via audio rather than via text. This AI voice
27:52 stuff is super cool. And in the tutorial section, I'm going to show you how to
27:55 take the exact same AI agent that we can chat to on a website and then connect it
27:58 to a phone number and talk to it on the phone. But then we get to what I call
28:02 automated agents. And so these are slightly different from the
28:04 conversational ones. The truth is that AI agents don't always need humans to
28:09 talk to them and use them directly. All they need is some kind of input or
28:13 instructions to trigger them and that tells them what to do. This means that
28:16 we can build these automated agents that instead of waiting for some kind of
28:20 human input, they are actually part of larger systems and processes and they're
28:24 triggered automatically by events like a new email received or a form submission
28:28 or they work on schedules like once a day and they essentially work in the
28:32 background without necessarily having human oversight or input. For example,
28:35 later in the video, we are going to be building an automated agent that is
28:39 triggered by a new form submission. When the form is submitted, some of that form
28:42 data is taken and sent to the agent, which then causes it to use the tools
28:46 that we've equipped it with and follows the instructions in the prompt that we
28:49 gave it in order to make decisions and take appropriate actions on our behalf
28:53 in a fully automated way. We are still sending the message to the agent, but
28:57 it's not a human needing to type it manually or speak it over the phone.
29:01 There's no human step. The input is being automated in some way. And this of
29:04 course opens up a huge number of use cases for AI agents in businesses
29:07 especially. And of course I'll be showing you how to build both types of
29:10 these conversational and automated agents in the tutorial section of this
29:14 video. But the last step of building your foundation of knowledge before we
29:17 move into that is to look at some real world examples of how businesses are
29:23 Real-World Applications
2:39 Why Learn to Build AI Agents?
2:41 in. Now, if you look at how long this video is, you'll realize that there is a
2:44 lot to cover here. And now, I don't want you to give up halfway through the
2:47 video. So, let's quickly take a moment just to get clear on why AI agents are
2:51 quite literally the next big thing. And I know that sounds cliche and you hear
2:54 it all the time, but seriously, they are. And why learning to build them is
2:58 by far one of the most valuable skills that anyone can have over the coming
3:02 decades. And if if my experience over the past 2 years is anything to go by,
3:05 that should be enough proof for you guys to believe me. So, stick with me. But
3:08 here's the hard truth about AI and jobs right now. According to the latest
3:12 research, McKenzie predicts that AI and agents could automate up to 50% of
3:16 current work by 2030. And the World Economic Forum reports that 41% of
3:20 companies plan to reduce their workforce due to AI. Now, a lot of this sounds
3:23 doom and gloom, and of course, many people are naturally worried about their
3:26 career in future based off seeing this kind of data. But it's not all bad if
3:30 you know where to look. And this video isn't about making you feel all sad.
3:33 It's about uplifting you. And if you look on the flip side of the same data,
3:37 these same reports reveal an enormous opportunity for those willing to seize
3:40 it. So the World Economic Forum's future of job report states that 50% of
3:44 employees plan to reorient their business in response to artificial
3:47 intelligence. And due to this reorientation, 66% of employees plan to
3:53 hire talent with specific AI skills such as prompt engineering. So on one hand,
3:57 we have the expectation of massive layoffs and automation of work over the
4:01 next 5 to 10 years. But on the other, we have the majority of employers searching
4:05 for people who have these AI skills or really just basic AI literacy. Why?
4:10 Because AI literate employees who can automate parts of their work can have 5
4:14 to 10x the output of someone who doesn't have any AI literacy. And I promise you
4:17 that brushing up on your AI and reaching this point of AI literacy so that you
4:21 can be on the winning side of this next 5 to 10 years is actually so much easier
4:25 than you think. I mean, it's easy as watching this entire video in order to
4:28 build your AI skills base to make a big step towards AI literacy. And if you
4:31 don't believe me when I say that a little bit of self-study like this video
4:35 goes a long way, here's a great clip that I've seen recently on the All-In
4:39 podcast from one of the most respected investors and technologists in the
4:42 world, Navar Raakan, alongside a whole bunch of other very smart billionaires.
4:46 Again, I would say the easiest way to see that AI is not taking jobs or
4:51 creating opportunities is go brush up on your AI. Learn a little bit, watch a few
4:55 videos, use the AI, tinker with it, and then go reapply for that job that
4:59 rejected you and watch how they pull you in. And so this video is exactly what
5:02 Naval is talking about. So whether you're an aspiring entrepreneur wanting
5:05 to learn valuable AI skills and launch an AI business like I have, or you're a
5:09 business owner wanting to just understand agents so you can use them to
5:13 grow your business, or maybe you're just wanting to make sure that you are the
5:16 last person that your boss thinks of firing because he or she's an AI wiz and
5:21 I can't afford to lose them. Then I have made this video for you guys. Now, what
5:23 I want you to do is close out all of your other tabs. Go get a notebook and a
5:27 pen and a beverage of choice and make sure you make a commitment to yourself
5:31 right now in order to finish this training and ensure that you are going
5:35 to be empowered by AI and not replaced by it. Now, if you've done all that,
5:39 let's get stuck into it. All right, so step one in building AI agents is knowing what the hell an
5:45 agent actually is. So, 2 years ago, when I first started learning about AI
5:49 agents, I had no idea what they actually were. The term AI agent gets thrown
5:52 around a lot in almost like everywhere these days. You got AI agents this, AI
5:57 agents that. But what actually is an AI agent? Well, the clearest definition
5:59 that I found that helps beginners to really wrap their head around what they
6:03 are is this. An AI agent is a digital worker that can understand instructions
6:08 and take actions in order to complete tasks. So, in a very simple way, just
6:12 like businesses have employees who handle different tasks, an AI agent is
6:17 like having a digital employee. But the cool thing is that you can build them
6:20 and you can make them do whatever you want. You're like literally building an
6:23 employee that you can put to work to do things for you. And of course, they cost
6:27 much less to run than a human and they don't need sick days and they don't
6:31 start beef with Mike over in the sales department because of his comment at the
6:34 coffee machine. So, I'm sure you can see the appeal of this kind of digital work
6:38 and AI agents to businesses who are looking to adopt them. In order to really understand why these
6:45 AI agents are such a big deal, we need to look at where we are coming from. So
6:48 most of you have probably encountered those chat bots on websites before. You
6:52 know those little little chat widgets that pop up saying like, "Hey, how can I
6:55 help you?" So these kinds of chat bots are pretty basic, right? They a lot of
6:58 the time they're useless and they're they're kind of like a waiter who can
7:02 only really recite the menu but can't actually take your order or or bring
7:05 your food. They can't do anything. They just respond with some kind of
7:08 pre-written answers. Well, nowadays it's a simple AI generated answer. But AI
7:13 agents are different, right? So, here's an example. If you ask a regular chatbot
7:17 about booking an appointment, it might say, "Oh, our business hours are 9 to5.
7:21 Please call to book." And that's it. They just give you some information
7:25 back. But with an AI agent, it could actually go and check the calendar, find
7:29 some available slots, go back and forth with the person that they're chatting to
7:32 in order to book an appointment, send you a confirmation email, then update
7:36 the business's scheduling system and CRM automatically in seconds. This ability
7:41 to take action is what makes agents so powerful. They're not just fancy chat
7:44 bots. They're actually digital workers who can search through databases, update
7:49 spreadsheets, send emails, book appointments, generate hold documents,
7:54 and much much more. And so building and deploying an AI agent is a bit like
7:58 hiring a new employee because when you bring someone into a business, you need
8:02 to firstly explain their roles and the responsibilities to them. You need to
8:06 give them access to your system so they can use them. And you need to trust them
8:10 to handle those tasks independently. And now when we are building agents, as we
8:14 see later, it's exactly the same, except these agents are going to be working
8:17 24/7. They're never going to get tired. They can be duplicated and modified
8:21 instantly. And they cost a fraction of what a human employee does. And this is
8:25 exactly why understanding how to build and sell AI agents is becoming such a
8:29 crucial and valuable skill these days. Because whether you're an entrepreneur
8:33 looking to scale your business or you're an employee wanting to become
8:36 irreplaceable and and make more money at work, knowing how to create and deploy
8:41 these digital workers is like the biggest cheat code in the whole world
8:44 right now. Now that you understand what AI agents actually are, let's look under
8:50 the hood and see how they actually work. Just like humans need a brain, memory,
8:55 and tools in order to do their job, AI agents need specific components in order
8:59 to function correctly. An AI agent needs five key parts in order to work.
9:03 Firstly, every AI agent needs a brain. In the AI world, we call this a large
9:07 language model or an LLM for short. And you've probably heard of some of these.
9:11 You've got GPT from OpenAI, Claude from Anthropic, Gemini from Google, etc. You
9:15 can think of the LLM as having a super smart intern who can understand your
9:19 instructions in plain English, and then figure out how to get things done from
9:22 those instructions. So, without this brain, all of the other parts would be
9:25 useless, right? It's like having a whole desk full of office supplies but having
9:28 no one sitting there in order to use them. Secondly, the brain needs
9:32 instructions on how to behave. And this is prompting. So writing a prompt for an
9:36 agent is how you program a lot of the behavior of it rather than having to
9:39 code it manually. And this is really what makes building AI agents so much
9:43 more accessible to non-coders as the way of actually programming the
9:46 functionality and how they work is done through clearly written instructions
9:49 rather than having to actually code it. Thirdly, agents need memory. Imagine
9:53 trying to have a conversation with someone who forgets everything you said
9:56 30 seconds ago, right? So, memory is really important because it allows your
10:00 agent to remember what you talked about just a few messages ago, keep track of
10:04 the tasks that it's been working on, build on previous conversations, and
10:08 even in more advanced ones, it can learn from your past interactions. And the
10:11 good news about memory is that most AI agent platforms completely handle this
10:14 memory component automatically. So, you don't need to worry too much about it.
10:17 But just know that it is an important part of a functioning AI agent. The
10:20 fourth component of an agent, and this one is optional, but it is external
10:25 knowledge. AI models like GPT and Gemini are pre-trained on a huge amount of
10:29 data, but that data is basically cut off at a certain point, eg 2024. It's kind
10:34 of like having a new employee who only really knows what they learned in
10:37 school. But just like you can train an employee like that with your company's
10:41 specific materials, you can also give an AI agent additional knowledge on top of
10:45 the information it was trained on through providing things like PDFs of
10:48 your company documents, spreadsheets with product information, customer
10:52 service transcript, or basically any other textbased information. Without
10:56 this added knowledge, agents will be limited to general information and
10:59 couldn't handle specific business tasks. But as I said, knowledge is optional and
11:02 you will only need it in some builds. Finally, and this is the most important
11:07 part, we have tools. So tools are what transform an AI agent from just being
11:11 able to chat to being able to actually get things done. So you can think of
11:14 tools like giving your digital employee access to different softwares. Just like
11:18 you might give a new hire access to your email or your calendar or your CRM
11:22 system, you can give an AI agent access to digital tools that let it take
11:26 actions when needed. These tools let your agent do things like checking
11:29 real-time data, updating databases, sending messages and notifications,
11:32 creating documents, all the stuff we went over just before and much, much
11:35 more. The really powerful part, which we're going to cover later, is when
11:38 agents use multiple tools together in order to solve complex problems, just
11:42 like us humans would use multiple different websites and softwares when
11:46 doing our tasks. Now, let me show you how all of these parts work together in
11:50 a real example. So, say you want an agent to handle customer support. When
11:54 the agent is sent a message, the brain immediately understands the prompt that
11:57 it has been given and also understands what the customer is asking. It checks
12:01 its recent memory before replying each time to understand the full context of
12:04 their conversation. And if the brand detects that the customer wants a
12:08 specific question answered from the knowledge base, it will use its external
12:12 knowledge in order to deliver the right information to them. And finally, it may
12:16 use tools to update a customer's account or to process a refund whenever required
12:20 during the conversation. So all of these things are happening in seconds as the
12:23 conversation is going on. Which is why AI agents are such a game changer. They
12:27 can combine all of these components in order to create a fully capable digital
12:31 worker that very very closely replicates how humans work. Now that you know the anatomy of
12:38 an agent and the five parts of it, a more practical framework for
12:41 understanding how we actually plan and build AI agents is what I call the three
12:45 ingredients. Basically, you only have three elements to plan when creating an
12:49 AI agent, which when mixed in various ways can create millions and different
12:53 types of agents for different use cases. This is because the AI model or brain
12:57 can be easily swapped in and out and isn't really a major factor in the
13:01 performance of the agent as any of the top models that you pick from any of the
13:04 different providers at any given time, they're all pretty good. And also, the
13:07 recent chat memory is handled by default in almost all cases when you're building
13:09 on these platforms that you're going to see later. What this leaves us with is
13:12 what really matters when building and planning AI agents. Firstly, the
13:15 knowledge, the external data that you want the agent to be able to use when
13:19 answering. Secondly, the tools, the different actions that you want the
13:22 agent to be able to take, eg saving the contact info to the CRM or getting some
13:27 live data on stocks or sending an email. And then finally, prompting, which is
13:31 the glue that ties everything together and determines how the agent behaves.
13:35 So, write these down. While the agent has five components, the brain, the
13:38 prompt, the memory, the knowledge, and tools, your main focus as an AI agent
13:42 builder is in the three ingredients of prompting, knowledge, and tools. In the
13:45 next chapter, we'll be looking at how you actually build an agent using the
13:48 different combination of these three ingredients. But first, we need to dive
13:52 deeper into the keystone of understanding how to build your own
13:55 valuable digital workers. And it all comes down to tools. Now, we need to dig a lot deeper
14:02 on tools as they are by far the most powerful part of AI. agents. But in
14:06 order to understand deeply and be able to build powerful agents with them, we
14:10 need to take a few steps back and actually cover the basics of how
14:14 software and the web and internet as a whole works. Now, this is as techy as
14:17 it's going to get in this video, but I promise once you understand this, it's
14:20 so important and it's literally like having a superpower. So, please stick
14:24 with me through this. So, remember how we said that tools are what allow agents
14:28 to take an action to actually do things rather than just chat? Well, the way
14:33 agents use tools and do work online is just how we do it as well, but with one
14:37 key difference. Instead of clicking buttons and typing into forms, agents
14:42 use what we call APIs. And every time you use the internet, you're actually
14:46 making dozens of requests to APIs as well and getting responses back, but you
14:49 just don't realize it. So, let me show you what I mean. So, when you click on
14:53 this video, here's what actually happened. Firstly, your browser sent a
14:59 request to YouTube servers saying, "Hey, I want to watch this video." And then
15:02 YouTube servers sent back all of the data needed. And thirdly, your browser
15:07 unpacked that data and started playing the video on your screen. So this
15:10 request and response pattern happens with almost everything that you do
15:13 online. When you open up Instagram, you are requesting your feed from Instagram
15:17 service. When you send a tweet, you are sending your data through Twitter's
15:20 service. And when you check your email, you are requesting from Google the
15:24 latest messages in your inbox and they're sending it back and your browser
15:27 is loading it. Thankfully, we get pretty websites and apps that make it very easy
15:31 for us to do this and use software via APIs through a nice application. But
15:34 under the hood, it is still two computers talking back and forth,
15:38 requesting, sending, and displaying new information for us on our screen. These
15:41 request and response happen through what we call APIs, which are application
15:45 programming interfaces. So, you can think of APIs like waiters in a
15:48 restaurant. Basically, they're going to take your order or your request to the
15:52 kitchen, which are the servers of the business, and then they bring back your
15:56 food, which is the response. So, you have request and response, and you have
15:59 you as the client, and them as the server. There are two main types of
16:02 requests that you can make. Firstly, either a get request. This is basically
16:07 just like asking for information like checking the weather or looking up the
16:10 price or loading this video. You're requesting to get the information to do
16:14 something. Secondly, we have post requests, which is when you're sending
16:17 some kind of information like posting a tweet, sending an email, or uploading a
16:21 photo. So, go back and write both those down because we're going to be using
16:24 them extensively in the building section of this video. Now, here's where it gets
16:28 interesting. So, AI agents use these same APIs as their buttons to do things.
16:32 So, each tool an agent has access to use is essentially an API that it is able to
16:36 call. So, these kinds of tools come in two different flavors. We have pre-made
16:40 integrations like Google Calendar or Gmail where it kind of comes out of the
16:43 box ready for you to use and just plug straight into your agent. And then we
16:47 have custommade tools that we can build ourselves. So you can think of pre-made
16:50 integrations like buying a readymade meal where they've done a lot of the
16:54 hard work versus custom tools where we are like cooking from scratch. And both
16:58 work, but custom tools give you a lot more control. And this is a skill that
17:01 I'm going to be teaching you in the second chapter of this video. Okay. So now you got the basics.
17:07 Let's get clear on how a tool is actually made and what the key parts are
17:11 as you're going to be using them a lot. So, let's break this down using a simple
17:15 example of a text capitalization tool. It takes in some text and the outputs
17:19 the capitalized version of it. So, first to create a tool, we need a function. We
17:23 need something that does work. In this case, it's super simple, right? It needs
17:26 to take in text and it needs to make it uppercase. So, this can either be done
17:30 through a basic Python function or you can use an LLM to do this as well.
17:33 Basically, we need to build some way to capitalize the text that we give to this
17:37 function and actually do the do the work. Next, in order for the AI agent to
17:41 use this function, we need to wrap it in an API. So, we have the function and
17:44 then the API wraps around it. And this is essentially making that functionality
17:49 we created accessible over the internet via APIs. Without it, the function
17:53 cannot be used by our agent. And in order to use the API that we've just
17:56 created and use this function inside it, the API is going to expect the same sort
18:01 of inputs that the function needs. So the input of the text that we want to
18:04 capitalize and it's going to output the capitalized version. So this is very
18:08 important to remember function. It takes in the input of the uncized text does
18:12 work and outputs the capitalized version. We're basically then just
18:16 building an API around it so that we can put it on the internet and then we can
18:19 have an agent that knows how to call that API can send information into the
18:23 input go through the function and then get spit out and then our agent catches
18:27 it at the end. But the magic step and what has really caused the AI revolution
18:31 to kick off is that we can explain to our agent how to use this API just by
18:36 explaining how the API works in natural language. And this is where schemas come
18:42 in. A schema is like a one-page instruction manual on how to use an API
18:47 and therefore how to access the functionality inside that API. And when
18:50 an AI agent is given one of these schemas, it too can read that
18:54 instruction manual and determine things like what the tool does, what
18:58 information it needs as an input, like we talked about before, and what
19:02 information to expect as an output. Now, they may look scary, but they're
19:04 actually really, really easy to understand, and we're going to cover
19:06 them in the next chapter with this video. And the good part about it is
19:09 that these days, schemas are automatically created by many of these
19:12 no code platforms that you build agents on. But I'm teaching you this because it
19:16 still helps to know what they are doing and what that what's really happening
19:19 under the hood on these platforms. And there are still going to be times where
19:21 you may need to roll up your sleeves and do it yourself. The incredible part
19:24 about these schemas is that modern AI like chatpt can read these instructions
19:28 and perfectly understand not just how to use it and like okay I need an input and
19:32 then I expect an output but also when to use it. For example, let's say we had an
19:36 agent and we gave it that capitalization tool that we just talked about and then
19:39 we said can you please capitalize this text? Mary had a little lamb. The agent
19:43 would then read over the schemas that we provided it and then it would see that
19:46 there's a tool with a description saying this tool capitalizes text right in the
19:50 instructions for the capitalization tool. We would have said this thing is
19:54 for capitalizing text and it takes in some text and it gives you the
19:57 capitalized version. And so the agent will read that and see okay this looks
20:00 like based off the instruction they just gave this is the tool that they want to
20:03 use. And then it will check the requirements and see that the tool takes
20:07 in one input in string format which is just text which we have described as the
20:11 text to be capitalized. So it reads all this. He says okay it it needs one
20:15 input. It's in string format. So I know I need to give it some text and okay
20:18 what does this text do? It's the text that they want to capitalize. Great. So
20:22 now it knows it needs the input and it knows that this is where it's going to
20:24 send the text to be capitalized. Then now that it knows what it wants, it goes
20:28 back to our message and it intelligently extracts Mary had a little lamp. not,
20:32 hey, can you please capitalize Mary had a little lamb? It's smart enough to know
20:35 that we want that taken out. So, it will take that part, Mary had a little lamb,
20:39 out of our input, and then it sends that to the API where our capitalization
20:43 function does its thing. Then the API sends back the capitalized version plus
20:48 a bunch of other response data as well. Then the agent looks at your original
20:52 question, looks at this messy response it got back from the API, and then using
20:56 its brain, the LLM, it writes a natural language response answering your
20:59 question. It would say, "Here's your capitalized text colon Mary had a little
21:03 lamb in all caps." That may sound complicated. It may have gone over your
21:06 head. Please go back and just listen to it again. You really, really need to
21:10 understand this process of uh the message comes in, looks at the schema,
21:13 realizes, okay, it wants to use this tool. Okay. What do I need to do in
21:16 order to use this tool? Okay. Well, then I'm going to grab it out of the input.
21:19 I'm going to put it in here. And it can actually go back and forth. Say our
21:22 capitalization tool needed some other input. Say you needed to provide uh the
21:26 number of letters you wanted to be capitalized. It may see that this tool
21:30 needs two inputs and I've only been given one. So then it will go back and
21:34 ask me, hey, could you can you please tell me how many letters you want to be
21:37 capitalized and you will see this magic in the agents that we're going to build.
21:40 When the agent can ask you questions in order to help fulfill the needs of the
21:43 tool, you have this very intelligent system that really will blow you away
21:46 when you see it in action. And one thing many people miss about this process is
21:50 the agent actually gets back raw computer data from the API or what we
21:55 call JSON. But using the LLM, it can transform that into natural conversation
21:59 and answer your question in a very very uh clear and concise way. So it's
22:02 basically like having an employee who can read all this technical information
22:05 and then explain it to you in plain English, which is another part of why AI
22:09 agents are so powerful. And so when you understand this pattern that we've just
22:11 gone through, I promise you, you will never see the internet the same way
22:15 again. Every action online is just requests and responses. And therefore,
22:20 we can build our own tools and AI agents to automate all of it. So instead of you
22:24 manually searching the web, copying information, pasting it into
22:28 spreadsheets, sending emails, an AI agent can do it all automatically using
22:31 tools if you build it correctly. It's like having a digital employee who can
22:36 press all of these API buttons for you thousands of times faster than any human
22:39 could. And don't worry if this feels a little bit technical. In the next
22:43 chapter, uh I'm going to show you how to create your own tools like this from
22:46 scratch using platforms like Relevance AI, uh where you can build out powerful
22:49 tools without writing any code. and will really start to click into place once
22:52 you see the stuff in action in the building section. But before we get into
22:55 that, let me reveal the power of AI agents which is unleashed when they are
23:00 given multiple tools to work with. Now, obviously having an AI agent
23:04 that just capitalizes text isn't very useful. I get that. The real magic
23:08 happens when you give agents multiple tools and the ability to use them
23:11 together in order to achieve complex goals. So, do you remember our
23:15 definition? AI agents are workers that can understand instructions and take
23:19 actions to complete tasks. When you give an AI agent a task, it's going to try
23:23 its best to execute on it, but if it doesn't have the right tools on hand to
23:26 do the job, it's going to be useless. And so, the more tools that you can give
23:30 an agent, the more flexibility it has to solve problems just like a human would.
23:33 So, let me give you a real example from my own business, right? Say I build an
23:37 agent and give it the task. Find AI startups that have recently raised money
23:41 and put them in a spreadsheet and add a summary of each of the businesses in the
23:44 spreadsheet and then email me the link to the spreadsheet. When you give an AI
23:47 agent a task like this and provide it with multiple tools to use, it can break
23:51 down this problem just like a human would. For example, it might think first
23:56 I need to search for AI startups using my web searching tool. Okay, let's do
23:59 that first. Then I'll need to create a new spreadsheet with my Google Sheets
24:03 tool. And then for each company that I find, I'll need to add a row to the
24:06 spreadsheet. And then I'll need to write a summary of each business and put it in
24:09 a new column. And then finally, I'll use my email tools in order to send the link
24:12 to Liam. And that's all great, but then when you add on top of that powerful
24:16 reasoning models like OpenAI's 01 and 03 and even things like deepseat as the
24:20 brain of the agent that can plan, take actions, then reflect and then plan
24:24 again and so on. You have essentially created a truly intelligent AI that
24:28 solves problems and approaches them just like a human would. So, say for example,
24:32 the original plan was to use the web search tool to search for AI startups
24:35 raising money. Probably a terrible search term, but what if that doesn't
24:39 return any good results to the agent? Well, a human would go, damn, I need to
24:43 change my search term or maybe I need to try find a different method of finding
24:46 these companies on like LinkedIn or something. The latest in AI technology
24:49 like these reasoning models, it allows these agents to do this exact same kind
24:54 of reflection and replanning in order to achieve their objective. And this is
24:57 when you can really see why we call them digital workers because they can do
25:01 things like planning multiple steps. They will use different tools in a
25:05 sequence and even adjust their approach based on the results from those tools.
25:08 Now, I should mention that this technology isn't perfect yet, right? So,
25:12 these multi-step tasks are often unreliable and agents typically need
25:16 human supervision for more complex workflows. But things are moving
25:21 incredibly, incredibly fast. In fact, we're already seeing the next evolution,
25:24 which is multiple agents working together. Instead of just one agent
25:27 trying to do everything, you can have one main agent that you give orders to,
25:30 and then it can use all of the other agents underneath it as tools where it
25:34 can send specific instructions. Like underneath the main agent might be a
25:38 research agent, which is best at finding companies and has its own tools. Then
25:41 you have a writing agent that's really good at writing summaries. Then you have
25:44 an emailing agent, which has got all the emailing tools. And so each of these
25:47 agents can be specialized in their specific task with multiple tools. and
25:50 then they all work together to achieve a common goal. This is exactly what major
25:55 companies like HubSpot and Microsoft and Google are building towards. It's these
25:59 entire workforces of AI agents that can handle complex business processes
26:02 automatically. In the next chapter, I'll show you how to build AI agents like
26:05 this for yourself using no code tools. But first, we need to understand the
26:08 different ways that these agents can actually be used in the real
26:14 world. So, we understand how AI agents and tools work under the hood. Now,
26:18 great. If you don't, please go back and take some notes, right? You should by
26:21 now have a whole bunch of notes um from the stuff that we've covered already.
26:23 And this stuff that you're learning took me two years in order to learn and and
26:27 be able to apply effectively. So, you best believe it that it's going to take
26:30 you two to three watches before it all sinks in. So, if you're feeling a bit
26:33 lost and and overwhelmed, don't worry. That's how it feels with learning
26:36 anything new or how it should feel if you're learning something that's
26:38 actually pushing your boundaries and adding something to your to your
26:41 capabilities. Next, we need to look at the different ways that AI agents can be
26:45 used in the real world. There are two main categories of AI agents.
26:49 Conversational agents and automated agents. Conversational agents are ones
26:52 that humans interact with directly through chat on things like websites.
26:56 You've got maybe you're chatting to it on WhatsApp. You've got interacting with
27:00 it over the phone via phone call. You've got chatting to it via Instagram DMs or
27:04 custom apps and websites. For example, OpenAI's GPT platforms allows you to
27:08 create agents that you can chat with directly on your computer or on your
27:11 phone. or using platforms like my own Agent, you can connect these agents that
27:15 you build onto a WhatsApp number or onto Instagram. And I'll show you how to do
27:19 this in the tutorial chapter of this video. So, in all these cases, you or
27:23 someone else is there sending messages or instructions to the agent and
27:26 explaining what you want to do and kind of chatting back and forth with it,
27:29 whether it's on a website, WhatsApp, Instagram, or whatever. And within these
27:32 conversational agents, it's not just text based. It's like I said, there's AI
27:35 voice agents as well, which are an extremely exciting sector of the AI
27:39 space right now. And these systems use multimodal models that can take in audio
27:44 as input and then produce audio as an output. And so these agents can be
27:48 chatted to over the phone or via audio rather than via text. This AI voice
27:52 stuff is super cool. And in the tutorial section, I'm going to show you how to
27:55 take the exact same AI agent that we can chat to on a website and then connect it
27:58 to a phone number and talk to it on the phone. But then we get to what I call
28:02 automated agents. And so these are slightly different from the
28:04 conversational ones. The truth is that AI agents don't always need humans to
28:09 talk to them and use them directly. All they need is some kind of input or
28:13 instructions to trigger them and that tells them what to do. This means that
28:16 we can build these automated agents that instead of waiting for some kind of
28:20 human input, they are actually part of larger systems and processes and they're
28:24 triggered automatically by events like a new email received or a form submission
28:28 or they work on schedules like once a day and they essentially work in the
28:32 background without necessarily having human oversight or input. For example,
28:35 later in the video, we are going to be building an automated agent that is
28:39 triggered by a new form submission. When the form is submitted, some of that form
28:42 data is taken and sent to the agent, which then causes it to use the tools
28:46 that we've equipped it with and follows the instructions in the prompt that we
28:49 gave it in order to make decisions and take appropriate actions on our behalf
28:53 in a fully automated way. We are still sending the message to the agent, but
28:57 it's not a human needing to type it manually or speak it over the phone.
29:01 There's no human step. The input is being automated in some way. And this of
29:04 course opens up a huge number of use cases for AI agents in businesses
29:07 especially. And of course I'll be showing you how to build both types of
29:10 these conversational and automated agents in the tutorial section of this
29:14 video. But the last step of building your foundation of knowledge before we
29:17 move into that is to look at some real world examples of how businesses are
29:23 using these AI agents right now. So firstly we have the personal
29:27 assistant category. And this is what most people think of when they hear the
29:30 word agent. something that you can chat to that's going to update your calendar
29:33 and sort of send emails and even make phone calls for you. Um, now these are
29:38 all nice to have features, but honestly uh this space is likely going to be
29:41 dominated by the big tech giants. You've got OpenAI through Chatbt trying to do
29:45 this with Tasks, Google through their suite of apps and connecting them to
29:49 Gemini and Apple through Siri. These guys are going to eat up this entire
29:52 market of personal assistance and your own personal AI agent that helps you do
29:56 personal stuff. the real opportunity lies in business applications and how
30:00 people like you and I can build and sell AI agents to businesses which we're
30:03 going to be covering in depth in the final chapter of this video. So, we've
30:05 got the next chapter which is going to be on building the four tutorials and
30:09 the final chapter is all about how to sell and how to monetize your AI agent
30:13 skills that you've just learned. One of the core use cases for businesses right
30:16 now are what's called co-pilots. And these are AI agents made for specific
30:20 roles in a business. We're going to be building one of these later in the
30:23 video. And these specialized AI agents are essentially helping someone in a
30:27 specific role in a business to do their job more effectively. Take a customer
30:31 support co-pilot for example. It would have a knowledge base that allows reps
30:34 to get answers to customer queries instantly and deliver them over the
30:36 phone. So they've got the little co-pilot up on the side there. They're
30:40 on the phone. They get a question, they can search and for an answer in the
30:42 knowledge base, it gives them back and they can give it to them over the phone.
30:45 This same agent could also have a tool that allows them to look up the customer
30:49 information very quickly. Um, I could have another tool that it makes it very
30:52 easy to send a summary of the call into the database so that the next rep who
30:55 picks up the phone and talks with them knows exactly what was discussed
30:58 previously. It's like giving every support rep some kind of AI assistant
31:01 that makes them dramatically more effective. It also makes their customer
31:05 support a lot more consistent as to what the company wants people to be saying,
31:08 which is a a big problem with managing large customer support systems. And then
31:11 we have lead generation and appointment setting agents. These are probably the
31:15 most valuable type right now. And businesses are using these on their
31:18 websites, through WhatsApp, on Instagram, and even over the phone to
31:21 engage and have conversations with the interested people who are approaching
31:25 the business 24/7. They can offer instant answers about products and
31:28 services. And they're even smart enough to be able to capture emails and phone
31:32 numbers mid-con conversation for later follow-up by sales team. Some can even
31:36 book appointments on the spot and mid- conversation by using a tool to check
31:39 the calendar availability and then using another tool to create a new booking
31:43 once they've agreed on a time with the prospect. Another real world agent use
31:46 case and one of my favorites is a research agent. And so these can help
31:50 businesses by automatically researching leads that come in through their website
31:53 or elsewhere. And when someone fills out a form, the agent can spring into action
31:56 and start searching the web for information on the company, finding
31:59 their LinkedIn profile of the person they're going to get on a call with and
32:02 gathering any other valuable data that it can find. Then it can take all of
32:05 this information and generate a summary of who this person is and what this
32:09 company is also and decide whether they're a good fit for working with the
32:13 company and if so then they can send the sales team some kind of detailed brief
32:17 or suggested strategy on how to close this particular person on a call based
32:20 on the research. So it's basically like having a an automated team of
32:23 researchers who as soon as leads show interest in your business, they're out
32:26 there figuring out everything about them and determining one whether they're a
32:29 good fit for you and your products and services which is called qualification.
32:32 Then secondly, if they are qualified, giving the sales rep something that will
32:35 bring them up to speed on who this person or who this company is and how
32:39 Foundations Summary
32:39 they can try to close them. So, we have covered a lot so far.
32:43 So, before we dive into each of these agent builds that I'm going to walk you
32:47 through over my shoulder, please make sure that you've got your notes taken
32:50 out and the core concepts of this video so far understood properly. You should
32:54 be clear on things like what is the definition of an AI agent? What are the
32:59 five parts of an agent? How is building an AI agent like being a chef? And how
33:02 many ingredients do you have to play with? What are the two main parts of a
33:08 tool? And what do schemers do? So, pause the video now and try to answer these
33:11 questions. And if you aren't 100% confident, you need to go back and watch
33:15 it again. So, don't rush this or you're going to feel way out of depth when we
33:18 get into the tutorials that we're going to be covering next. But if you are,
33:21 congratulations. You are one step closer to AI literacy and becoming a much more
33:26 valuable uh participant in this global economy. So before we get into the
33:28 second chapter, there's just three very quick things from me. Firstly, if you
33:32 are a business owner who wants to fast track to becoming an AI leader within
33:36 your industry, at my agency, Morningside AI, we offer everything from AI
33:39 education and upskilling programs for executives and staff to AI strategy and
33:44 roadmap consulting and of course AI development services as well. So we
33:47 would love to help you get ahead. So feel free to get in touch via our
33:50 website in the description below. And secondly, at Morningside, we are hiring
33:53 for all sorts of roles right now. So whether you want to build AI systems for
33:56 some of the world's biggest companies that we have as clients or to help
34:00 produce videos like these that are seen by millions of people or create
34:04 educational material for thousands of businesses. Uh we have roles for all
34:06 sorts of things right now. So you can apply using the link in the description.
34:09 And please, even if you're just vaguely interested, I really recommend you just
34:12 check out the link and see what roles we're hiring for. Uh you never know
34:16 what's going to be on there. Um and it may be a very good way for you to use
34:18 your skills to fast track into the AI space by working under myself and my
34:22 team. And finally, if you have gotten any value so far in this video, please
34:25 head down below and leave a like on the video. It helps me reach more people.
34:28 Um, I put a lot of work into these videos and it also lets me know that you
34:31 enjoy this kind of content and that I should make more of it. And of course,
34:33 if you like this kind of content and want to see more of it, you can
34:36 subscribe so that YouTube will put my videos up for you whenever a new one is
34:39 released. So, there's also a little share button if you want to click that.
34:41 That'll let YouTube know this is good content and that you're sharing it to
34:44 other people. Not only will that help me, but you can share it to your friends
34:48 and family who may also or you may want to help them to brush up on these skills
34:51 or help them give a way to get on the front foot with AI. And that's what I
34:54 really make these videos for. So, thank you for sitting through that little bit
34:57 of housekeeping and self-promotion. Now, building. I have carefully assembled
35:00 What We’re Building
35:05 this chapter on building to give you the most bang for your buck possible in
35:09 order to kick off your AI agent learning journey. We are going to be covering
35:12 four different use cases across four different AI agent building platforms.
35:16 These are all no code, so don't worry about that. And the chances of you
35:19 falling in love with at least one of these platforms is pretty much 100% as
35:23 you're going to rapidly start to connect the dots uh about how you can start to
35:27 use these kinds of agents and these platforms in your own life or in your
35:30 work or for your friends and family and those around you. So, here's a quick
35:33 rundown of the builds we're going to be getting into. The first build is going
35:36 to be a sales co-pilot built with relevance AI. And here we're going to be
35:40 building three custom research tools from scratch, including an advanced web
35:44 scraping tool, which is a a great skill that I want to teach you. And with
35:46 these, we are going to be creating a conversational agent to help the sales
35:51 reps at Big Boy Recruits, a hypothetical fantasy uh recruitment firm, in order
35:55 for them to be better prepared for sales course. So that's the purpose of the
35:58 sales co-pilot. The second build is going to be an automated lead
36:01 qualification agent. And this will be built on a platform called N8N. And this
36:05 time we will be helping Big Boy Recruits, our fantasy recruitment firm,
36:09 to automatically research and qualify new leads and then send an email
36:12 notification to the correct sales rep. And this is going to show you that
36:15 automated style of agent where it's built into a process rather than having
36:18 a human input necessarily. In build number three, we will be building a
36:22 website and phone-based lead generation customer support agent. This will be
36:25 built on voice flow and it's going to be able to do three things. Firstly, answer
36:28 questions from a knowledge base, generate instant quotes using a custom
36:33 tool we build and also do lead capture on interested prospects. We're then
36:36 going to slap this agent onto a website widget so that you can chat to it via a
36:40 website and via chat and text. And then we're going to take that exact same
36:43 agent and connect them to a phone line so that we can call our agent over the
36:45 phone and access all of the same functionality we just talked about. And
36:49 finally, for build number four, we'll be using my own AI agent platform, Agent,
36:53 to rapidly build a lead generation agent and connect it to a WhatsApp number that
36:56 we can chat to. The leads that we collect are going to be automatically
36:59 sent into an Air Table database for later review. And please don't skip
37:02 around these builds as they're all kind of connected in some way where we're
37:05 reusing parts from build one and build two, etc. But without further ado, let's
37:09 get into building some agents. All right, people. Enough of the theory. Uh,
37:14 now we get into the fun bit of actually building these agents out. So, I've done
37:18 a lot of work and my team has done a lot of work. So, thank you to the my team
37:20 members who have helped me put this together. Um, putting together four
37:25 different AI agent builds for you. And this is really going to walk you through
37:28 an A to Z all the different platforms that you really need to care about, all
37:32 the different kind of core use cases and functionality. There's a lot more of
37:35 course, but this is going to really give you the foundation that you need to
37:39 succeed in the space. And hopefully it'll be the thing that kind of sparks
37:43 your interest in it because I I want you guys to have fun with it. these big
37:46 tutorials for me. Honestly, when I put a lot of work into it, I build up the sort
37:49 of mental resistance to it because I know how much work there is going into
37:52 it and I have to make this big whole session where I'm all uptight about it.
37:55 But I'm just going to try and relax and enjoy this. And I really want you all to
37:58 do the same. So, set a bit of time aside. You can either pause this video,
38:02 put on your watch later, but I really want you to take your time with this.
38:06 I'm going to be doing this more. So, when you do tutorials like this, there's
38:09 a few different ways you can do it. I can either do all the building and then
38:12 give you the templates and kind of just spoon feed it to you. And that's more so
38:15 what you do for someone if you're trying to like really fast track them and they
38:18 don't want to learn all the skills, but um I I know what I'm trying to build
38:21 here for you guys. And I'm going to give you a sort of stream of consciousness.
38:25 You just get to see me kind of jamming out and building these things. And I'll
38:28 be explaining my thought process and the concepts etc along the way to reinforce
38:33 what we've learned before. So I'm just going to dive into it with our first
38:34 Build 1
38:40 And so what I've done is put together a big Figma board here which is going to
38:42 be breaking down all these different builds. So under here there's some
38:45 goodies you see. Oh, there's some goodies under each of these that I've
38:48 put together. Um and we're going to go through them one by one. Starting off
38:51 with agent one over here. I mean there's a lot of stuff here um that you guys are
38:54 going to get. So you'll get the whole Figma and it includes all the templates.
38:57 So if you do want to just kind of watch through this, pick it up. You can either
39:00 do it and follow it step by step with me and see how I build it and really build
39:03 those flexible skills that you're going to need to succeed in the space or you
39:07 can just watch it and be like, "Okay, I kind of get what he's doing and then
39:10 take all the templates from me at the end." That's I mean, completely up to
39:13 you. Depends if you want to be a really really nerdy builder about it and get
39:17 into the weed like like I like to do. Um or you just want to be like, "Hey, I
39:19 want to do this my business. I want to roughly understand how these things work
39:23 and what platforms." So, use this resource as you will. But we're going to
39:26 jump into agent build number one here, which is our sales co-pilot built with
39:30 relevance AI. So, running through this quickly, we have the purpose of this.
39:33 This is basically going to look a bit like this. It's going to be a co-pilot
39:37 and co-pilots work in that you have a uh it's basically a specific AI agent that
39:42 you build for a specific staff or staff member or role. So, say this case, it's
39:46 going to be a sales co-pilot. It'll be the thing that the sales rep uses to uh
39:50 in their day-to-day as they're working on their jobs. You can add tools like in
39:54 this case, you see we're going to have three different tools here for our
39:56 agent. One's going to be a company researcher tool. So this is when the
40:00 sales rep would be like, hey, I have a call coming up soon. Um, let's put in
40:03 this I need to research this company cuz this is who I'm going to be on a call
40:06 with. So they'll put in the company URL. This tool that we're going to create is
40:08 going to go and research that company. It's going to bring back and give a
40:11 summary. And then it's like, okay, well this is the LinkedIn URL of the person
40:14 that we're going to be got on a call with shortly. It's going to pass in the
40:17 LinkedIn URL. It's going to take that URL. It's going to pull all the
40:20 information and write a summary about the person. So now we have the company
40:23 summary and we have the person summary. And the final step here is going to be
40:26 what I'm calling a pre-all report generator. And that's going to take both
40:29 that company and prospect research that we've done. It's going to combine them
40:34 together and be for this specific company. As you're going to see that
40:37 this hypothetical company we're building this sales co-pilot for, it's going to
40:41 generate a basically a pre-core report or a strategy uh a strategy prep for the
40:44 sales rep so that they go onto those calls much more prepared and also sort
40:49 of a personalized guide on how to try to close this person. So, um, all of these
40:53 templates are going to be here. Each of these are templates for the tools. And
40:57 this is for the agent as well. Um, but here's some more information. You guys
40:59 can pick through this as you wish. But, um, this is the kind of end result and
41:02 we're going to be able to chat to it. And this would be something you could
41:04 build for a client. You could build it for your own business or you could just
41:07 tinker around. You could build co-pilots like this on relevance for yourself. So,
41:10 that's why I want to start with relevance because it also is a platform
41:13 that we can build these tools on. So, it's a really, really good one to start
41:16 with and let's get into it. So, the first step of course is to go to
41:19 relevance AI. So, I'll put a little link up here. You guys will be able to get
41:22 this Figma. It'll be on the school. Um, all of the information and all the
41:24 resources for this are going to be like this Figma is going to be linked to the
41:27 school. My free school community. If you haven't already joined, biggest AI
41:31 community on school, biggest AI business community probably in the whole world.
41:35 Um, so we can jump across that first link in the description. You'll be able
41:37 to find this in the YouTube resources section. Um, pretty straightforward. Of
41:41 course, when you click on this, it's going to ask you to log into relevance.
41:44 So, if you haven't already, you can make an account. Um, it's fairly low cost.
41:48 They have a free plan, then a team's plan, I believe. Um, so it's not too
41:50 much, but it is a really, really valuable tool as you're going to see.
41:54 So, you can sign in here. I'm going to jump in with my Google account. There
41:58 may be a bit of setup for your account, but I'm sure you guys are smart enough
42:01 to figure out how to set up an account. I'm sure relevance also makes it easy
42:04 enough. So, then we get taken to this dashboard, but we see on this left hand
42:07 side, we have tools. So these are the tools that I've talked about um where
42:10 it's some kind of functionality that we can create and we can build it all on
42:13 relevance no code and there's even sort of extensibility or you can add more
42:18 functionality in to relevance by adding some low code components or even custom
42:21 code. So relevance is a really great base for building not only tools but
42:24 then the agents that you can connect that into and we're going to use the
42:27 same relevance tools that we make now in multiple of the different agents that
42:29 I'm going to make for the rest of this video. So first things first if we go
42:32 back to the Figma here we see we need to make three different tools. So tool one
42:35 is company researcher. It's going to take in a company URL. It's going to
42:37 search the web and it's going to return a summary. So, that's the functionality
42:42 we need. Let's go and create a new tool. Going to call this um
42:48 research company. We can give it a cool little I'm going to zoom this up for you
42:51 guys. Hopefully, that's the right size. Um uh search. Have fun with this stuff,
42:57 guys. Like, if if it's putting stupid emojis on things to to enjoy it, then
43:01 then that's what you need to do. Like, uh if you make it a chore, it's going to
43:04 feel like a chore, right? Um, now we get to descriptions. Um, this is something
43:07 we're going to see recurring, but basically as as you know, as we learned
43:11 in the concept section, we need to have natural language descriptions of our
43:15 tools and of our APIs so that the agents can read those descriptions and
43:19 understand uh what the agent or what the tool does and what those different parts
43:22 does. So, you'll see this recurring throughout this. But first things first,
43:26 what does this research company tool do? Um, takes and oh, I got caps lock on.
43:32 takes um and there's also some tutorials here if you want to go deeper. Relevance
43:36 has some great documentation as well. Um but takes in a company
43:45 penny URL and scrapes the website then returns a sum and AI generated summary. So then to
43:57 build a tool we need to have some kind of input. You don't always need an
43:59 input. It can actually just be triggering, but generally you're going
44:02 to have some kind of input that the agent needs to pass into the tool. In
44:05 this case, to research company, it's going to need to take in some text,
44:09 which is going to be a URL. Um, we're going to say comp company
44:15 URL. And then again, here we have another description. You see, describe
44:18 how to fill this input. This is again going to help our agent within relevance
44:22 AI and elsewhere as you'll see in tutorial number four. Why this is so
44:25 important to add in the descriptions, right? So this is a URL for a company to be
44:39 researched must be in the format https colon slash um dot dot dot dot dot. So we need to
44:48 have the https for this to work. So that's going to be our input. Now if
44:50 this seems a bit confusing just stick with me. It will make sense in a second.
44:53 this stuff. If there's anything that I've learned from picking up so many
44:56 different tools, like when I first got into Facebook ads, when I've got into
44:59 building these kinds of agents, it's you feel completely overwhelmed, but that's
45:02 all just part of the process. And what feels difficult now is not going to feel
45:05 difficult forever. So, just please stick with me. Um, and it's a really, really
45:09 great feeling once you go and be like, this was hard a few weeks ago and now
45:12 it's really easy. So, we've got our first input. That's what the research
45:15 company tool is going to do. And then we need to define our steps. So, the next
45:19 step, we can go add. I'm going to hide this so we get a bit more space. Add
45:22 step. Now, the cool thing about relevance is that it comes with a lot of
45:25 great functionality out of the box. Here's one, extract website content, we
45:30 have LLMs, we have Google searching, we have all sorts of AI generations,
45:35 replicate um knowledge bases as well. There's so much cool stuff on here and
45:38 this is why I really rate relevance as one of the best platforms. If you were
45:42 to go all in and want to upskill, you can build so much on this. Um, so I'm a
45:45 big fan. I love the the relevance team and what they're doing. So the initial
45:48 plan for this build I was going to use the extract website content which is
45:52 fairly straightforward. We can say oh one other thing um the company URL we
45:56 are going to be using this company URL throughout our tool here. So we can
46:00 change this text to say something more descriptive. So company
46:06 u URL. You guys are going to got going to think like think and write like
46:09 coders now cuz you need to uh use some kind of syntax and use some kind of
46:13 variable naming convention. This is a a standard one or you can do things like
46:18 company URL camel case but I prefer this format as I'm I'm mainly a Python kind
46:21 of guy myself. Now that we have that named we can use it in these kinds of
46:25 fields. So here you can see pops up use inputs. So basically when this tool is
46:29 run it's going to take the inputs or the information we put in the inputs and
46:32 it's going to pass it to different steps and use them as we describe within the
46:35 within the builder here. So let's run morningings.ai. Um, and then we can
46:49 click uh run here. So, that's going to go to my website and scrape the
46:54 information off of it. There we go. So, it's got all of this, but you see it's
46:59 just pulling back the first page. Um, and this is why I actually I shuffled
47:02 this around. And I want to show you guys how to do something a bit more advanced.
47:04 I know this is supposed to be a beginner tutorial, but this is not really that
47:08 useful and I it's a very easy thing for me to just bump this up to a little bit
47:12 more valuable. Um, while relevance tool here is great, we can do better. So,
47:16 we're actually going to delete this. Um, you can use that step for all sorts of
47:20 other things, but I really like what's called fire crawl firecroll web scraper. This is a cool
47:29 app. Um, firecrawl.dev. Shout out to the guys at firecraw. Basically, if we then
47:33 go HTT Oh, I should just see if they can doti and now do a free scrape for us
47:42 here. So, this is just going to do the single URL just like we got in
47:46 relevance. But the difference here is if we then go to crawl, if you hover over
47:51 this, it's going to crawl a URL and all of its accessible subpages outputting
47:55 the content from each page. So, instead of just taking that front page, it's
47:57 actually going to crawl through multiple things. So this is really the first cool
48:01 thing or or first skill that I want to put in your tool belt is that you have
48:04 things like fire call that you can use their relevance. They have things like
48:06 map which is just going to output all the URLs that it finds. Then there's
48:10 other things here where you can use AI to extract data. I'm not going to go
48:12 into that. But what we want is this crawling functionality from firecraw. So
48:17 I'll put a link in on the school post that comes with this video. It again
48:19 first link in the description to go to the school and if you go to the YouTube
48:23 resources section um there will be a uh a whole post on this and all the
48:26 resources will be in there. It'll also be in my free course on school as well.
48:29 So you can find it in the classroom section. So you want to sign up to FCL
48:34 so you can get API key. It's very easy. We can just go through with Google.
48:37 Again, this is not sponsored and there is zero sponsoring going on through any
48:40 of these tools. I guess I'm kind of sponsoring my own tool because I'm
48:43 putting it at the end. But I'm not getting paid a dime for any of this. I'm
48:45 really just trying to put you guys on what I like to use, what's made me
48:49 money, what's made me a more valuable AI automation expert or developer for my
48:53 companies and for the companies you work with. continue and then you get to the
48:57 dashboard here. It might look a bit scary, but what you can do is go to the
49:00 API keys. So, you can click on create an API key here, YouTube. I'm having issues with mine. It
49:06 should be fine for you. I've already got an API key, but once you get the API
49:09 key, what you can do is take it and come back here to relevance. And you can go into the side
49:16 panel here. Where is it? Settings. And then we have our API keys. So, we're
49:19 going to need to add more into this later. So, keep an eye on this. This is
49:22 something you need to be familiar with. Um, and you can scroll down to firecrawl
49:25 and you just pop it into this firecall API key section here and you're good to
49:29 go. And you can come back. Oh no, we don't want to duplicate
49:35 that. Now we've got our firewall set up. We need to make a couple things. We need
49:39 to do a couple tweaks here. We have, of course, the if you want to get those
49:43 variables up, you can go bracket bracket um or curly bracket curly bracket, which
49:46 is shift um to get those. I don't know why it's not popping up. There we go.
49:51 Company URL. And if we hover over this, we can see it says scrape the provided
49:54 URL only. Uncheck if we want to crawl instead. So if we want to get that crawl
49:58 functionality that we just saw that we think we want to get all the data, we
50:02 can uh uncheck it. And then we want to extract the main content. So you might have to just trust
50:07 me on that one that we don't want to have all of the other rubbish. We just
50:11 want the body of the website. Um a number of pages. We don't want this to
50:13 take too long. You can expand this much more. Um but I'm just going to go for
50:18 say five for now just to keep it uh keep it quick. And then now what we can do is run this
50:24 again. We've still got my URL up here. We can go run step. Give it a second. You will at some
50:31 point have to pay firecrol. Of course, it's not a free service, but they do
50:33 have a free plan, so you should have no issues with getting that. So here you
50:35 can see we're getting a lot more data back from this web scrape than we were
50:39 with just the relevance version, which is great. So the next step is we have
50:42 this data. We want to generate some kind of research summary um so that we can
50:46 send that to our sales script once they have requested it. So now it gets into
50:51 the fun part of writing LLM prompts. For this one, sometimes you need to really
50:55 go and and make a big effort, which we are going to do later as you'll see. But
50:57 in this case, we just want a quick summary. I'm going to throw one in here
51:00 that I made earlier. All of this will be available on the Figma or it'll be given
51:03 somewhere in the resources, right? But um if we just need something quick and
51:06 dirty here, it's not really a massive part of the project, so it's okay to
51:09 just whack one in there. So bang, I've got it in there. Can you please take
51:12 this website content and summarize it into a 300word natural language summary,
51:15 which clearly outlines rad where they're based, their values, etc. anything that
51:19 would be helpful to know for a sales rep who will soon be on a call with them. Uh
51:22 break it into key areas like overview, products and services team etc. And I've
51:25 got a couple things here to make sure that it doesn't mess up which I it was
51:29 doing for me a bit in testing. So we do want to put in if we go curly bracket
51:32 curly bracket we want to put in the fire call data here which is going to be uh
51:36 all the website data that came back from our scrape. We want to then insert it
51:39 into this prompt here. So I hope you're starting to the things in your your cogs
51:42 and your brain are starting to click into gear here. And then we get to
51:45 select the model for this. It's pretty basic task. So, I want something quick
51:48 and cheap. Um, for many of you, it's going to be easiest to work with the
51:51 Open AI APIs because you've probably already played around with your API key
51:54 before, which I'll show you how to do in a second. But, let's go with GPT40 Mini,
51:58 but Relevance, of course, does have support for all of uh the other models.
52:02 But my tendency for most tasks now is to actually go for some of the Google
52:05 models that have come out. Again, you guys might be watching this in a year or
52:08 whatever. It might be quite different, but at the moment, Google is really
52:11 leading the way with making the cheapest models possible, and they're actually
52:14 really good as well. the the price decrease on using things like uh Google
52:17 Gemini Flash Light and Google Gemini Flash 2.0 and stuff like that. It's
52:21 ridiculously cheap and it's also a really good model. So, it's a bit more
52:24 difficult to get APIs on the Google side. So, I'm just going to stick with
52:27 OpenAI for this tutorial. Um, so let's go GPT4 mini. And then we need to of
52:31 course set up our API key. Um, if you scroll down, where is the OpenAI? So, to
52:39 platform.openai.com.playground/playgroundplayground, sorry. Um, this will again be linked in
52:43 the resources or you can just type up platform.openai.com. I'm sure you'll
52:46 find it. If you haven't already, create an account um, and sign in. And then you
52:50 need to go to your dashboard here. You go on the left side, you go to API keys,
52:54 and you click create a new secret key. And then you're going to be able to copy
52:57 that key and bring it back into uh, relevance. Paste it in here. And then
53:00 you're good to go and start using the OpenAI suite of models. It's pretty
53:04 easy, right? So now we have our information back from the scrape. We
53:09 have our prompt in here. And then we can go run step and see what this company
53:12 research tool is going to output for us. Boom. Morningside AI is a leading
53:15 artificial intelligence development company dedicated to empowering
53:18 businesses autonomous AI agent development, enterprise consulting,
53:23 chatbot development team. Uh they got keep me out of my own team page.
53:27 Damn. But uh yeah, so there you go. That's the that's the company research
53:31 tool. That's step one. Um I hope you guys can kind of see how that works. Now
53:33 a cool thing about relevance is there is this build section which we've just gone
53:36 through. But you can also go to use and this can be really helpful when sharing
53:39 these kinds of tools. So not this is not really only just an AI agent tutorial.
53:42 I'm also teaching you how to build tools because you can build very valuable
53:45 tools and something like relevance. And then you can go share. You can go
53:49 publicly available. Oh, and I can click this and then I can give to my employees. I can
53:56 give to the companies that I'm working with, the clients that I've sold these
53:59 kind of services on. And then they get a nice and handy tool like this. I mean, I
54:02 use these throughout my organization for like description generation, a lot of
54:06 content repurposing. There's tons and tons of different use cases for building
54:08 these kinds of tools. And then you can take this URL and you can share it
54:11 around to whoever you want. So, this is a uh a great way to use relevance. Let's
54:15 go back to our tool here. But for now, we need to keep moving along so we can
54:18 get this this first agent done. So, that's the first tool. I'm going to run
54:20 through it a bit quicker now that we know how these kind of tool buildings
54:24 work. I'm going to create a new tool here. I'm going to call it this time if
54:28 we go back to our Figma. And I recommend when you guys are building your agents,
54:30 you're building systems and planning them out. This kind of laying out in a
54:34 Figma, if you're not familiar, this is Figma. It's like a design software, but
54:36 you can also use it for kind of whiteboarding. It's called a Fig Jam
54:39 board. It's one of the types of uh boards you can do. And I use this all
54:43 the time with my team as well. It's like if you can't take what I'm telling you,
54:46 lay it out on a board so that I can review it and give you notes and and
54:50 then we all agree this is the build. Um there's often a lot of uh communication
54:53 issues with explaining functionality. So this AI agent layout, it's really
54:57 helpful. You can if it's maybe a workflow automation, you can do box box
55:01 arrow arrow arrow all of this laying out how it's going to be built. Here I've
55:04 done it for you in a very basic format. So we've done this first one. Maybe I
55:07 just make this green. Um second one is going to be prospect researcher. So this
55:11 takes in a LinkedIn URL, searches the web, and returns a summary. And you're
55:14 going to see how I tie this all into an agent shortly. So we're going to go research
55:22 prospect. Sil takes in a linked in URL scrapes the profile and then generates
55:28 an of the prospect input. We're going to need a link and URL the link linked and
55:35 URL of the prospect. this. Now, we're going to add a step and
55:46 relevance has got us here linked um get a LinkedIn profile or company
55:53 post. So, this is cool cuz then we can pop in our LinkedIn URL
56:03 them. So, we're going to get the user mine my LinkedIn profile, if you guys
56:12 want to connect with me on LinkedIn, more than welcome to do so. I'll put in
56:17 the description below. Um, we can do a little run step here. So, if we go back
56:22 over to data here, that's great. We've got my about section. So, this goes very
56:26 long way across because it's in uh it's in JSON here. It's got my company. It's
56:33 got my company domain, where I'm from, years, company, founded, tons and tons
56:36 of great information that you guys can use and we are going to use shortly. So,
56:39 this is really cool. I would probably add one more step to this if I was
56:42 taking this um and building it for for my own team. I would add in another
56:47 LinkedIn scrape here um where we just do the same thing, but we also get the
56:50 posts because posts can give you a bit more up to date um information on what
56:53 they've been doing recently. that you can guys can add that and you just go
56:57 add a step LinkedIn and you do the same thing as we've done here but you change
57:02 this to LinkedIn post get user post so that may be a cool thing for you guys to
57:05 add the uh functionality on at the end is a bit of a challenge you can pause
57:07 this video and do that and then we need a llm step to take this again I'm going
57:12 to grab a pre-written prompt that I did just to save some time going to drop
57:17 this in here says fairly similar stuff um LinkedIn data I'm going to put all
57:23 the data in there. And then we're going to use a GPT4 mini again. And we can give that a
57:30 run because we've already got this data queued up here. And there we go. We've
57:34 got a nice summary. Um, if I change this to nice and formatted. So there's a
57:38 little button down there between raw or formatted. And you can copy the stuff
57:41 out of here, of course. So Liam Mley, my followers. Damn, I got a lot more than I
57:44 thought. So there we have the summary, my name, where I'm based, uh,
57:48 information, my career experience. Super handy stuff. And this is going to be
57:51 super helpful in the next step when we generate that pre-core report. So very
57:55 quickly, we've created one more tool. I'm going to save this. So now we've
57:59 got, if we go back to our Figma, we have two of these done. Now the final one is
58:03 going to take in the company and prospect research and generate a
58:06 pre-call report. So this one's going to be a little bit different. If I go
58:11 create a new tool, preall report tool. Okay, free call takes in company
58:21 and prospect summary and generates a free call report for sales
58:26 direct. And now for the inputs for this, we need long text, not just normal text
58:29 because we're going to be taking in that big company and uh and prospect
58:34 research. So we go prospect summary summary of the prospect based on length
58:43 that and a prospect in this case is someone who's a potential customer. Just
58:47 to clarify that if you're new to business and don't really get these
58:58 summary, right? We have our prospect summary and company summary in there. I
59:01 hope you're following along. Next step is just an LLM step and we want to
59:04 combine these two together. Again, I've got a little handy prompt for this to
59:07 save us time. Now, in this case, you will see that the prompt is a bit
59:11 bigger, right? So this is um for more important parts when you are creating
59:14 tools whether it's it's for agents or just generally when you're using prompt
59:17 engineering and LLMs to create value. In this case we are creating value because
59:21 in this case we're taking in this prospect summary and this company
59:24 summary. We're also giving it the the context of this fantasy or or
59:28 hypothetical business that we are selling this agent to as like a co-pilot
59:32 system. We've got Big Boy Recruits which is Dallas based recruitment firm
59:35 specializing in software industry talent acquisition for SMBs. You're going to
59:38 see this kind of recur across the different projects we do, but basically
59:41 we're helping these big boy recruits to automate their business with AI. So,
59:45 this takes in some context on that business and it's going to generate a a
59:48 report that's going to help the sales rep say, "Okay, this is the company.
59:51 This is what we sell. This is what we specialize in. This is the company that
59:54 we're trying to sell to. This is who we're going to be talking to. What's
59:57 some how can I personalize this call or what's the strategy I can go into this
60:00 with? What are some angles that I can attack this call from?" And so to do
60:04 this, I have a prompt writing tool that I use quite regularly and my team uses
60:09 it as well. Um, perfect prompt. So, this tool does a lot of leg
60:13 work for myself and my team all the time. I'm going to give it to you guys
60:15 to use for free. You'll be able to clone it into your relevance account.
60:18 Basically, what I'll usually do is I'll put on uh the dictation thing. You've
60:22 got it on one of these keyboards. You've got this little thing. basically
60:26 whatever on your computer allows you to speak into the computer and it takes in
60:30 your voice and transcribes it into into text on the screen. I'll press that and
60:33 then I'll explain as you can see here what is this prompt doing and why and
60:36 I'll go this prompt needs to do this this this is going to take in this
60:39 information it's going to do this the reason we're doing this is this this and
60:43 I'll do like a big big body of text in there and then the next one if I have
60:46 them I'll give some good examples of input and output pairs of how I want it
60:49 to take in data and how I want it to spit it out. If you give it both of
60:52 these and you hit run, it's going to print you out using the researchbacked
60:55 prompting techniques that we use at Morningside. It's all crammed in here.
60:59 There's a video that I recommend all of you watch. It's going to be in the free
61:01 course anyway on school. So, when you get in there and watch my prompt
61:04 engineering guide, um, this is basically the entire information of that prompt
61:08 engineering guide smashed into this LLM step here. So, when you pass this
61:10 information in, it applies all of that and it gives you out a prompt that is
61:14 fully researched back and performs very, very well right out of the box. So,
61:16 that's a little bit of extra value I wanted to throw in there for you guys.
61:19 this is going to be available on the school um with the rest of the resources
61:22 as well. So basically I put in the information here about what this
61:26 particular uh task was. I said it's going to take in the prospect
61:28 information. It's going to take in the prospect summary and the company
61:31 information and it spat out this basically first go and I just had to
61:34 insert these variables. So this prompt and everything else will be on the on
61:37 the score resources as well. So in this case because we are doing a bit of
61:40 strategy and sort of high level thinking rather than just summarizing and we may
61:43 want to change the model here to something a bit smarter. We could go to
61:47 03 mini which is one of the later ones. Um, again, when you're watching this, it
61:50 might be 06 or 010 or whatever the hell they come up with next, but there's
61:53 probably going to be some much better models. So, just use a smart one because
61:57 it's really strategizing on how big boy recruits can position themselves for
62:00 this call. So, enough of me yapping about that. Let me grab some inputs for
62:06 [Music] it. Shout out muscle. All right. So, I've got this information here about
62:12 myself, my LinkedIn profile, and my company. And so again, remember that
62:15 this is for Big Boy Recruits, a Dallas recruitment firm specialized in software
62:18 industry talent. So it's going to look at my company, Morningside AI. It's
62:21 going to look at me and my background and my profile on LinkedIn. Then it's
62:25 going to spin, as we see here, um, review this, analyze this, map big boys
62:29 unique value proposition, ra, and it's going to try and create a report that's
62:33 going to allow the sales rep to sell me or close me better on there or find some
62:37 angles to sell to at least. So if we go run tool, and now you see that I'm using
62:40 a lot of just basic web scraping and LLM steps. I just want to show you guys the
62:43 basics. The thing is tools can get very very advanced when you have like CRM you
62:47 want to integrate into, but relevance allows you to do all of that. It's just
62:50 within the scope of tutorial, it can be pretty difficult to be pulling
62:52 information from all over the here cuz I have to set up a database, show you guys
62:56 how to do it, too. So, this this keeps it quite confined, but it still gives
62:59 you a good taste of it. So, if we look at this view, all key business
63:03 challenges and opportunity, Morningside AI, this Liam's profile gives me a bit
63:07 of a rundown of this mapping big boy recruits unique value proposition. Maybe
63:10 it's going to be better if I change this to the format. There we go. Talks about
63:15 mapping big boys recruits, strategic talking points. I've been following your
63:20 journey of digital marketing, AI, ra um dive into opportunity. I work at big
63:23 boys. This assist, and it's even gone and done a section on anticipated
63:26 objections. So, the idea is that the sales rep is going to have a skimmer of
63:29 this before the call, which ties into the value that I've listed here, which
63:32 I'm going to do for all of these builds, by the way, which comes down to
63:35 ultimately a better prepared sales rep should close more deals, right? if they
63:38 know more about the prospect and the company and you have an angle to try
63:42 sell through or suggestions at least. It should increase the conversion rate of
63:45 the sales team. So, we've built this tool. We can change this to green now.
63:48 And the final step is going to be heading over to our agent builder within
63:52 relevance. I'm going to save this. If you pop over in the left panel here, you
63:56 can go into our agents. And what we want to do is create a new
64:03 agent. We're going to call it our sales co-pilot. Um, big boy big boy sales co-pilot. Sorry, I
64:12 got, like I said, I got to have fun with the stuff where I go kind of insane. Um,
64:18 this this agent is our sales co-pilot that helps reps to be better prepared for sales
64:30 [Music] calls. Triggers, we don't need to do any of that. We go to core instructions. I'm
64:35 going to again paste in some of the stuff that I've prepped earlier. So, if
64:37 I paste this in here, you'll see that it's structured fairly similarly to the
64:41 prompt that we just did before for the uh pre-core report generator. And this
64:45 is again using another tool that I've created um for AI agent prompting. Um
64:49 so, it's fairly similar stuff that I I include in that other prompting tool,
64:53 but the agents is slightly different um to just regular LLM steps within
64:56 different tools and workflow automations. So, I will include this as
65:00 well. It's my AI agent perfect prompt generator and it's fairly
65:02 straightforward to use. I'll include that in there as well. But basically,
65:05 you put in all the information about what the agent is, why it's doing it,
65:07 the different tools that you're connecting to it, and then it prints out
65:10 this for you. So, I'll just run through it. We've got a role here telling it who
65:13 it is, um, and kind of hyping it up and saying how good it is. Explains the
65:17 task, um, talking about how it's helping to conduct detailed research on
65:20 companies and prospects. Um, some specifics. Uh, don't need to worry about
65:23 those too much. Just reiterating the task. And now here we can enter in the
65:27 references to tools. So if we go slash tool and see in order to get the agent
65:31 to function as well as possible, we need to tell it what tools it has available.
65:35 This is really key across all the agents you build, especially if they're more
65:38 conversational. You need to explain to them what tools they have and how and
65:41 when they should use them. So if I go, company. Um, yep, that's right. Purpose,
65:50 input, company URL, use when needing to gather company information. This one of
66:01 prospect. And then this last one is our [Music] pre. There we go. So that's all whacked
66:05 in there. Don't need to worry about that too much. But again, this will be
66:08 included um this prompt and everything if you want to follow along and also if
66:11 you just want to clone the whole agent um and use it in your own business or
66:14 sell it or whatever you want to do. We also get to select the model here. Um
66:18 I'm just going to keep it as GPT4 mini. I like some pretty fast responses here
66:21 because agents as someone's using it, it can feel really irritating if it's not
66:25 responding quickly. So, we've got all that built out. That's the core
66:29 instructions in the prompt of the agent. Um, we can go down to the uh tools
66:33 section. It's got all the tools connected in here because we mentioned
66:36 them in the prompt. And we can just go through and do some quick settings on
66:39 here. Um, I don't want to have to do an approval for it. Some tools you can say,
66:43 look, they've got to give it a thumbs up before it can actually uh trigger it.
66:46 Um, prompt for how to use. Just some quick descriptions we can pop in here.
66:49 I'm not sure why relevance doesn't carry that over from the tool. I guess they're
66:52 asking us to do it again for some reason. Um when you need to research a
66:55 uh prospect linked and URL, we're going to say this is auto run as well. Um and
67:05 then preall change this to auto run as well. Use this when you need to generate a pre
67:15 call report from the company and research. All righty. Um, and there's
67:21 all sorts of other cool stuff. Relevance, as you can see, is like
67:25 abilities, sub aents, metadata, extra stuff that you can build onto. Um, but I
67:27 just want to get you guys started with the core of this. Um, all right. So, we
67:36 that. And boom, we have our agent here ready to go. So this is where you can
67:38 test your agents and use them if you want to. But in this case, I'm just
67:42 going to give it a quick rundown and say see if the functionality is working as
67:47 we as we planned. Um, hi, I am getting on a call with Liam Otley from morning
67:55 side AI. Here uh has lenol report please to prep for the call. Sending your task to big boy sales
68:24 copilot. And then we get to see all the debug and how it's actually walking
68:28 through these different steps. Oh, let's does. Yep. Okay, that's great. It's
68:36 using the research company as we wanted it to. Should add in one more step there to
68:42 research the prospect as well. There we go. Using the second
68:46 tool. It is pretty satisfying when this stuff works. And this is just a really
68:49 basic one, guys. I don't want you to think this is like, oh, well, that's
68:52 pretty underwhelming, Liam. I'm trying to teach you the basics so that you can
68:56 actually build on top of this. So if you get the bug, if you get the like you get
69:00 a travel bug, if you get the agent bug and you see the stuff and you're really
69:03 interested. Oh, look, there it is. Now it's filling out the prospect summary as
69:06 the inputs. Surely we don't have to watch it do that. It's going to take a
69:11 while. [Music] Um, I really write that out word for word like that. But when you start to
69:18 see this magic and you add in other cool tools and functionality, you test it on
69:21 yourself. You can build like things for maybe you want to do content, you make
69:25 yourself a little content co-pilot, etc. There we go. To use all the tools
69:29 and it should be spitting back and bam, there we go. So, I hope that was worth
69:33 the wait. Let's go through it now. Here's your comprehensive precore report
69:35 for your upcoming conversation with Liam Mley from Morningside AI. Pre-core
69:39 report. Talent acquisition under pressure. Morningside AI operates in a
69:42 highly competitive AI and tech market as they scale. Finding specialized talent,
69:46 engineers, data scientists, AI with a proven record can be challenging. That's
69:50 scarily accurate because that is literally one of the biggest constraints
69:53 that we have had to scaling Morningside long-term is that it's just really
69:57 really hard even with my channel. It's so hard to find the right people and get
70:00 them to commit as as developers as well. So if you want to build a very big
70:03 general AI development firm, an AI automation agency, you need the best
70:06 talent and you need to get a lot of it in um so that you can scale up. So
70:10 that's bloody spot on. Obviously this thing knows that's a good angle to sell
70:15 through. Um but yeah, prospect analysis. Liam's a dynamic entrepreneur and
70:17 thought leader with a robust background in e-commerce, digital marketing and AI.
70:22 His journey reflects a passion for innovation and commitment to continuous
70:26 learning. Again, that's pretty bloody spot on um hands-on experience. So there
70:31 you go. That is the big boy sales co-pilot for big boy recruits. The cool
70:34 thing you can do now once you have built this um is you can go share um there's a
70:40 chat UI which I'm going to turn on now. There are chat widgets so you can put
70:43 them on websites and stuff like that. What I want to do is just pull this up
70:46 because this is what you'd be giving to your client likely or if you guys are
70:49 going to start selling these to businesses which again we're touching on
70:52 selling in the last section of this video. So how do you turn these into
70:55 into a business and start making money from it and selling these as a as a
70:58 service and building these businesses which is really where the money's money
71:01 is made. So there you go. This URL you can obviously send to your client. If
71:03 you're building it for your own team, you can send this to your team and say,
71:06 "Hey, pin this because you're going to be able to use it. Add more
71:09 functionality into it, etc. That is how All righty, that is build number one out
71:20 of the way and we are jumping into AI agent build number two, which if you're
71:23 listening closely at the start of the section, we are talking about an
71:28 NATbased inbound lead qualification agent that's going to be doing some
71:31 pretty cool stuff for us, which is a really important function within a
71:34 business um around lead qualification. Um, so this is a really cool one. Again,
71:37 in this case, this is what we call an automated agent, not a conversational
71:40 one. What we just built is a conversational agent. humans are
71:43 directly talking to it and chatting back and forth and using it and we are
71:46 operating it ourselves. In this case, as you can see on this little flow uh flow
71:49 diagram here, this is a screenshot from the final product. We are actually
71:53 baking this into a workflow that's going to be triggered on a form submission.
71:56 We're going to do some research and then we're going to use the handy AI agent
72:00 and tools agent within NA10 to trigger another workflow and then send some
72:04 emails off. So, this capability of using AI agents in workflow automation really
72:08 expands the possibilities of what you can build. And the software NAN that I'm
72:11 going to teach you how to use is really at the the cutting edge and leading the
72:14 charge when it comes to these automated uh AI agent workflows. And just quickly,
72:17 it's good that we walk through the purpose and the value behind this
72:19 automation so that we know why we're doing it. Right? So this inbound lead
72:23 qualification use case is based on the fact that companies who market
72:25 themselves well soon have far too many people reaching out to them. Many of
72:28 which are not a good fit or what you'd call qualified for what they sell. Eg
72:32 they're too small or they're not the right industry. Like you you have a
72:36 business and they say we only help XYZ kinds of businesses. And if leads come
72:38 to that business who are not qualified, then they obviously don't want to be
72:41 taking calls or or doing anything further with them. So this process of
72:44 researching a new lead and deciding whether or not to take a call is known
72:48 as qualification, which is what this agent aims to automate. So the value
72:51 here is that instead of having to pay someone to manually qualify and go
72:54 through all of these leads or using arbitrary rules, which is what some
72:57 businesses have to go to, it's like look, oh look, we've got so many leads.
73:00 Let's just say if they don't want if they say that they're not on this, then
73:03 we'll just cut them out. And that's potentially leaving money on the table
73:05 by cutting out leads who would have actually been a good deal, but the rules
73:09 kind of didn't see enough detail to be able to determine if they're a good fit
73:12 or not. So, this automation is essentially immediately qualifying and
73:15 triggering the next steps to the sales team and allowing them to do that human
73:19 style research on these leads at scale. So, enough talking. Here's a little bit
73:22 more information. Again, with all of these, it's going to be on the figure on
73:24 the school. Then, I've also broken down how this agent would actually operate in
73:28 the real world and sort of real time. A lead's going to be submitting the form,
73:32 which is going to be this here. um the relevance company AI researcher. So
73:34 something that we've just built in relevance, we're going to reuse in here,
73:37 which is handy that you can start to move these components around and see how
73:39 they can fit into different automation platforms. Um then the AI agent is going
73:43 to look at this information from the research. Then it's going to determine
73:46 based on a qualification criteria we give them um inside the prompt of this
73:51 agent whether they are qualified or not. If they are qualified, it's going to uh
73:55 use this tool here and call this second workflow. when we're going to
73:58 essentially analyze that further and do a notification to either our agency team
74:03 or our our SAS team and if they are not qualified we will use this tool here
74:06 which will just send an email straight back to the person who submitted the
74:09 form say hey sorry we're not open to working with businesses like yourself at
74:12 the moment let us know if we can help you any other way so that's a rough
74:15 rundown of the build let's jump into it so to kick things off of course you need
74:20 a platform on natn.io io. All links and resources will of course be on the school uh post for this
74:24 video. And once you're on this page, you can go to get started and you can create
74:29 an account for free and just go through the setup process that they do. I'm sure
74:32 you can figure that out. They do have a 14-day free trial as of this filming. So
74:35 that's very good if you're just jumping in, not having to pay anything. And they
74:37 give you quite a lot of usage up here. As you can see, 1,000 executions. So
74:40 what we're going to do, of course, is click on create workflow up here. I will
74:43 be giving you the template. So if you want to just import it, you can. That
74:46 will be on the Figma there. But just like the last tutorial, I'm actually
74:49 going to be showing you the process of building these up from scratch that you
74:52 can see how I how you go through the process of building these automations
74:56 and the the testing and back and forth you need to do in order to get to the
74:58 end result. That's probably actually a lot more important if you are to go out
75:01 if I'm trying to teach you to fish, not just give you a fish, is to see how I
75:04 deal with problems when we're building these. So, let's get started by starting
75:09 off our trigger. We're going to go form has a nice form on new inn form event. And here we
75:17 get to create an N8N form. So we can call it uh work with your. So now we get to pick the field
75:31 names. So we want to have the first one is make that a required field. We add
75:40 another one. What is your company website? field. Right. So, we've just built out a
75:55 basic form here with first name, company website, which we're going to need for
75:58 the next step. I've put a placeholder in here so they know that it needs to have
76:02 https um at the front of it. And I'm asking for them to provide some
76:05 information about your inquiry like and maybe you can say what can we help
76:10 you? Um and that's going to be text area. So, they've got a bit more room.
76:12 So, we can go test step here. Make sure that it's all looking nice. This is what
76:15 the form's going to look like. What's Obby. There we go. We've submitted that
76:24 form. If you go back, there we go. We have the data in. So, this shows you the
76:28 output. NAM works by having kind of this middle island here, which is what you
76:31 set up. And then the left side is the input and the right side is the output.
76:34 So, we've tested it and these are the outputs that our form is giving, which
76:38 is what we are looking for. And the next step in this lead qualification process
76:42 is to do some research on the lead. So we have the company URL and this is when
76:46 we're going to go and make an HTTP request. So this is basically calling
76:50 any API over the internet. And in order to set this up, we actually need to go
76:57 relevance and we're going to find the company researcher tool that we made
76:59 with relevance. And now you're going to start to see how this all fits together.
77:03 um that building tools and relevance can also be very useful and an extremely
77:07 useful skill in all areas of AI automation because now I can come on to
77:11 research company and not only can I use it here this I mean this is why I think
77:15 relevance is such a great platform um you have the use here so I can send this
77:18 across to a client I can share it with it as I showed before I can use it just
77:22 here myself I can run it in bulk on a spreadsheet or but more importantly in
77:27 this case I can go to the API and now I can call this might look scary just
77:31 don't worry I can call this functionality basically send in a
77:35 company URL and get back the research. I can access this over the internet
77:38 through an API and they give me it here and they tell me exactly how to call it.
77:41 So now we this is on actually a post request. So we can copy this. Remember
77:46 how we talked about get and post request. This is a post request because
77:49 we're posting some data to relevant AI. So we change this method to post. We put
77:54 the URL in here. We do not need authentication in this case but you can
77:57 turn it on. So you can make a private here and then you can have an
77:59 authenticated. might sound a bit complicated but for now don't worry
78:02 about it we don't need to have an authentication step on this and then if we look at the request
78:09 body here it tells us how we can send data to relevant AI and if we go copy
78:16 here come back we're going to send a JSON and we can change this to using
78:26 JSON and then we can paste in basically what we have been given from relevant AI
78:29 now this looks a messy. Let's pop this open a bit more. And we actually need to change it
78:35 to an expression here. So fixed means that we're not accepting any dynamic
78:39 data in. So when the form is submitted, we actually need to take in some data
78:42 from that form which we have over here. We need to inject it into this uh HTTP
78:47 request to relevant AI to get this company research done. So we can't have
78:51 it as a fixed um JSON body here. Need to change it to expression and then we can
78:54 pop this out here and it gets a bit easier. So, we have params. We have the
78:59 company URL. And then we need to pop in here. Oh, the company
79:05 URL. Pop that in there. And here on the right side, we get to see what that
79:08 would look like given the test data that we've just put through. So, you can see
79:12 I've got the company URL, https morningside.ai in quotations here, which
79:15 is what we want. So, we can go back now. And now we can give it a test to see if
79:18 it's going to be able to communicate with relevant AI and get us the data we
79:24 want. There we go. We have our result back from relevant AI which is the
79:28 summary. As you can see when we go back to this um and back to build this is
79:33 exactly what we'd expect. You know you put in the URL does this scrape a fire
79:37 call writes the summary spits out the summary and this summary is what we're
79:40 getting back out over here which is what we want. So bank that's great another
79:43 step done. And those of you who are a bit confused about what this is this is just
79:48 the body of the request. So because we are sending a post request remember how
79:51 we have get and post request. Get requests are typically just with a URL
79:54 and with a bunch of stuff tacked on. A post request, we need to send a a JSON
79:59 body like this. And it does look quite confusing, but if I take this and go
80:09 form, paste this in and beautify it. And you can see how we have basically the
80:12 project which is the relevant project that we are calling. And this tells the
80:15 API this is the project that we want to interact with. and the project expects
80:20 the params, the inputs, which is the company URL, and we're injecting that
80:25 company URL from our information here in NA10 that we've dragged across. It might
80:28 seem tricky a few times, but trust me, this stuff becomes like riding a bike
80:31 once you get up and running. So, um, a few more of these and you'll be you'll
80:34 be completely fine. All right. And so now that we have our company
80:36 information, the next step is going to be setting up the agent, which is really
80:39 the coolest part in my opinion about make right now is that we come in here
80:44 and we can click on um agent and we can set it to as a tools agent, which means
80:48 we connect our own tools. And if we just back out of this, we can do cool things like set up
80:53 the model. And right here, this is so cool because we get to see exactly what
80:57 we were just talking about earlier in this video, but we have the different
81:00 parts of an agent, the different ingredients, right? So here's the chat
81:03 model. This is the brain. This is the LLM that's going to be powering the
81:06 whole agent using the soup example. Like this is the one meat that we get to
81:10 choose, right? This is a specific model. In this case, we going to be using
81:13 OpenAI again because you already have your API key and I can't really be
81:15 bothered going and showing you a whole another provider, but it's the same
81:18 process for all of them. You can if you want anthropic, you can then pick all
81:22 the anthropic models. Uh but in this case, just to keep it simple, let's just
81:30 to and open AI model. And if we go back again, you can see that we now have uh the memory and the
81:35 tools that we can connect. So, of course, we have the tools, which we've
81:38 talked about a lot in this video already. We can connect multiple
81:40 different tools here, as we're going to do in a second. And then we also have
81:43 memory set up here, which is a little bit outside of the scope of this video.
81:46 As I said, in most platforms, it comes builtin, but NAT is a bit more of a
81:49 developercentric platform. So, if you wanted to play around with memory, um
81:52 different forms of of managing memory, you can do that here. In our case, we're
81:55 not going to be touching it. And in order to connect a knowledge base, if we
81:58 wanted to set up a knowledge base for our agent, we would connect it as a
82:01 tool. So you can see here we've got in-memory vector store, pine cone vector
82:05 stores, etc. These are vector databases that we can connect just like in the
82:08 other tutorials we're going to do in this video. When you upload some
82:10 documents to make a knowledge base, they're essentially being put in a
82:13 vector store like these, but the platforms manage it for you and make it
82:16 a lot easier to do. So in this case, we're just going to be doing two
82:19 different tools. Firstly, we're going to be calling an NATM workflow. So I'm just
82:23 going to finish off the the basic setup of this. Um, and then we're going to add
82:26 another tool on. It's going to be the Gmail tool. And then before you know it, we
82:33 have our AI agent structure built out. So, we're using the Open AI models.
82:36 We're going to pick the model shortly. We're going to be using the tool to call
82:39 the second NATM workflow, which is going to be uh triggering the the email
82:43 notifications for our sales reps and the classification of the uh of the lead.
82:46 That's going to make a bit more sense when we actually do it. So, just stick
82:49 with me on this. And then this Gmail is going to be sending back a hey sorry you
82:55 didn't qualify for what we do um sorry we can't help you let us know if we can
82:58 do anything else for so to start setting things up we can start from left to
83:01 right here with the chat model um the openai model that we want to use and you
83:04 need to set up your openai account so you can click create new credential and
83:07 you need to go and add in your API key here I'd suggest you go and make a new
83:11 one on platform openai.com and you can add a new one in here and you can name
83:14 the key nat so you start to know which keys are used for which different
83:18 platforms um and once you put that in there you can Just click save and then
83:20 it will run a little test and then you're ready to go. You should have this
83:23 set up here. Then we get to select the model that we want to use for our agent
83:26 here. And in this case, I want to go for something quite smart. So, I'm going to
83:28 go for 03. We have 03 mini here. This one appears to be a little bit more recent.
83:34 So, they'll sometimes put the dates on the end of it. And if you just look at
83:36 the current date, you can kind of see how how close to the current date it is.
83:39 Um, but I'd say this one's a bit more recent, so it's going to be hopefully a
83:42 bit better. And next, we're going to skip this cuz we need to set up a whole
83:45 different workflow to connect it to. Um, we're going to jump straight into this
83:47 Gmail one. You need to set up again another connection as you go through all
83:50 of these different automation platforms. You do need to do these these
83:54 connections between uh your own account, say your Gmail or your calendar or all
83:57 these different apps. You need to go and create a new credential. Um and you can
84:01 just do the sign in with Google here. Super easy to do. I'm sure you don't
84:04 need my help with that. Once you've set up that connection, you can close this
84:07 and you will see the connection that you set up here. Basically, that's what
84:09 we're going to be sending emails through. And then we can get into
84:12 setting this up. So, because this Gmail tool is going to be used to send a a
84:15 reply back to the person who submitted the form and say, "Hey, sorry, you're
84:18 not a good fit for us." You want to send it back to the person who submitted the
84:21 form. Now, if you scroll down here, uh, yep, you see that I've I've forgotten to
84:25 add the email form in. So, this is a good example of needing to go back a
84:28 little bit. So, we can go back to the form submission here. Um, scroll down,
84:35 say, what is your email? And we can set it as an email. So, it's going to automatically force
84:41 them to provide a valid email for us. And I want to maybe shimmy this up a
84:49 Um, so it's name, email, then company, website. Um, we can do another test here
84:55 just to give it some proper data. Oh, it's not going to let us do
84:58 that because this isn't set up properly. So, we can just delete that for
85:03 now. And we're actually going to delete that. Otherwise, it's going to be a bit
85:06 of a pain. again. So, we can submit another form here and go back to NA10. And there we
85:20 go. We have the information. We have the email now. That's great. And so, now we
85:23 can come and set up our tools again. We've got the workflow there and we've
85:28 got the Gmail. Um, and now we have our Oh, and we haven't got the data here
85:41 because we need to run this again and get the research. So, now the
85:44 research is done in there for us to set up the Gmail tool here. We can go to two
85:49 and we'll be able to pull in the email. So, again, they submit this form. We
85:52 realize that they're not a qualified uh person for our offer. And then we're
85:55 going to send an email back and say, "Hey, sorry, you're not a good fit." So,
85:59 we can say the subject here is um thanks for your interest. I'm going to change
86:04 the email type to text here and I'm going to write a basic message in. I'm
86:07 just going to snag it from the one that I've done previously. So, we need to
86:10 change this from fix to an expression because we want to be pulling in their
86:13 name here. So, I'm going to paste in what I have here. Again, this would be
86:16 included um in the resources. It's just going to save us time if I don't have to
86:19 type this all out manually. Um but you can see here, I'll just delete this so
86:23 you guys know what we're doing. If I go hi, or hi, and then we can add in what
86:29 is your first name? Hi, name. In this case, it's going to be filling in my
86:32 name here as an example. So, highly mly, thank you for your interest in big boy
86:35 recruitment services as you specialize in recruitment for software and
86:38 development agencies. We're not a good fit based on your company's industry.
86:40 Please let us know if you'd like to connect us with one of your partners who
86:43 specializes in dealing with your needs. Cheers. Huge Jackman here to sales, big
86:48 boy recruits, BBR, uh, Dallas, Texas. So, if you guys remember huge Jackman,
86:52 comment down below um for the OG fans. And then that is our Gmail all set up.
86:56 And just quickly so that you've got a bit more knowledge around how this Gmail
86:59 uh tool works, we have all these different steps that we can use. We're
87:02 using the send one. So it's sending an email. You can use reply, you can use
87:05 get, delete, all these other functions, but the easiest one and the most common
87:08 one you're going to use is going to be that send one, of course. Now, we need
87:11 to set up this NATM workflow which the agent is going to call as a tool um when
87:15 they are a qualified prospect. So I'm going to delete this one and just save
87:18 this for now. Then we're going to go back to home. We're going to create a new
87:25 workflow. And this is a really cool skill that I want to teach you. The fact that you can
87:29 build all these workflows and then connect them to agents and it can just
87:32 be taking data in and kind of shooting it off in all directions and triggering
87:36 all these complex multi-step processes because it's a super valuable way of
87:39 using agents. Um, so I really want to teach you that. And obviously this one's
87:42 going to be starting off a bit different to the other one. We actually want this
87:45 to be set up as when executed by another workflow. So that's going to be what the
87:49 trigger is here. And we are going to be able to define using fields below. Let's
87:54 just add in one here that is a lead lead name. Um, for now we can just leave that
87:59 there. But that's all that we need to set up. We need to go back over to the
88:01 other one. Just needed to set this part up. Let's rename this qualified lead
88:07 lead classifier and notifier. So we can save that there. And
88:10 we have a bit more work to do on this other one. If we go back to this, we can
88:16 rename this here. So let's call this our lead qualification agent.
88:23 And so now we have our other workflow set up. We can come here. We can call
88:27 another workflow with the tool and we can call it lead is qualified. And so
88:31 now is when we get to tie back into what we learned in the foundation section
88:34 because we are now writing descriptions for our tools. Remember how we had
88:38 schemas and scheas are basically written instructions or instruction
88:42 manuals on how to use tools and how to use the APIs that wrap around them. Um
88:45 this description is going to be basically those descriptions that you
88:48 put in the schema. But NAT is going to be basically constructing it for us in
88:52 the back end. And we just get to put in here, okay, what's the name of this
88:55 tool? It's going to be called lead is qualified. It's giving us a nice example
88:58 of how we can write a description for the tool. So call this tool to get a
89:00 random color. The input should be a string with a comma separated names of
89:03 colors to execute. So in our case, we can say call this tool when the lead is
89:07 qualified according to our criteria. The inputs should be lead name, lead email,
89:18 company, company summary, and request in info. We're basically telling that AI
89:22 model or the brain what this tool can do. So when we send it some kind of
89:25 input, it then it looks over our tools. It looks at the Gmail description and it
89:29 looks at this uh workflow tool description and goes, "Oh, well, I have
89:31 a tool that does this and a tool that does this. What have they just sent me?
89:34 Okay, now I think I know what I need to do from here." So this is the rough gist
89:37 of what we want to do as a description for this tool. I'm actually going to
89:40 beef it up with a bit of a a bigger one here. Um if the lead is qualified to
89:44 work with big boy recruits, eg they are software based business like SAS or
89:47 development agencies and trigger this tool and send the lead data in the
89:50 following format. It's just dummy data. So name a name email um an email message
89:57 I want new div qualified true company information and company information
90:01 which is a summary of the relevance tool to do the company research that we have.
90:04 So, might seem a bit crazy at the moment, but stick with me because it
90:07 will make sense in a second. We're basically just told it when it's going
90:09 to trigger this tool and the format to send the data in. And then for the
90:13 workflow, we get to choose here the one that we just set up, which is our
90:17 qualified lead classifier and notifier, at least in my case. And then we see the
90:20 workflow inputs that we've just set up. So, if we go back over to our other
90:24 automation, so when we open this up and define our inputs here, you can see over
90:28 here we are getting just one of them that we've put in as an example so far.
90:32 So now we need to set up all these inputs correctly. And we have the name
90:35 that we want. So we've got the name and message. Um, honestly don't think we need this
90:54 qualified one here. And then we have the So, if we just test this, head back over
91:13 we refresh this list. Oh, back out. Save it. And this pops up. It says that these
91:21 inputs are outdated. So, there we go. We have lead name, email, message, etc. And
91:24 then we can actually automatically fill out a lot of these inputs. Maybe I will
91:27 put this back in here just to show you qualified. Um and then we just call it
91:39 uh true And we go back here and we add in one more which is a
91:58 qualified which is also a string. And so we have this qualified field here as
92:02 well. If we go back um I'll just test this. Save it again. Come back over and update
92:09 this. I think we can actually make it even cooler. So let's go. Um it changes
92:13 from a string to a boolean. So that's either true or false. Um, and if we test
92:20 this, save it again, and we change this to take away these little things. Sorry,
92:24 pisses me off if I don't have this set up right. Um, and then we update this,
92:28 you'll see this turns into a switch. So, that's a true or false, right? In order
92:31 to trigger this tool, it should be qualified by default. So, it's a little
92:34 bit redundant, but it's still cool to show you how we can get AI to fill out
92:37 these fields. Um, in a lot of cases, we don't, but for this qualified run, we
92:41 can. So, we can set this as an AI generated field. We can say if the elite
92:47 is qualified based on our criteria this set to true. And then for the rest of
93:07 quickly. Give this a second. Open this back up. And we can fill out some of these
93:14 fields. So we can go lead name that we want to pass through to the other
93:17 automation is going to be that. So we can fill a lot of these in email
93:26 um message and the company information um we can get from um this technically
93:30 but um maybe we could do a cool AI generated one here which is um a short
93:37 summary of the company and the industry that they are in. company's details. So, this is basically
93:46 telling the AI model of our agent how to fill this field out, which is one of the
93:49 reasons that AI agents are so powerful. So, that's all set up. Now, we've got
93:52 all of our tools set up. The last step is just to set up a prompt for our
93:56 agent. And I am going to cheat a little bit here and just throw one in that I've
93:59 done before to save us a bit of time here. So, we want to set the prompt
94:05 here. We can paste in this information here. And this is basically just telling
94:08 the agent who it is and what it's supposed to do. You're a lead
94:11 qualification agent. Your job is to analyze the form submission and company
94:14 research provided and then decide whether they are qualified to work with
94:18 big boy recruits. Ra we specialize in XYZ. Um we are specialist in capturing
94:22 talent for ra. We only work with softwarebased businesses, EG SAS
94:25 companies or development agencies. These companies are willing to pay much more
94:29 developers than your average marketing company or local business. Therefore, we
94:31 only work with them. Your job is to determine if the lead you are provided
94:35 with is a good fit for big boy recruits. And if so, call the lead is qualified
94:38 tool and send the elite information to it. If lead is not qualified, then you
94:41 must trigger the Gmail send email tool for us to respond to them letting know
94:44 letting them know we are unable to work with them. And then we have a response
94:47 format here which we can probably just delete. And then we can add in here is
94:50 the lead to information for you to analyze. Let's pop this out to make it a
95:07 We can add in um just go name um company URL company and we take it. So that's from
95:22 the relevant step for the research that we did to scrape using our firecrawl
95:25 tool. And then we provided all the information to this agent and it's going
95:29 to be injected with all of these values on each form submission and then it's
95:32 going to make a call on what tool it needs to use. So we're pretty much
95:37 there. We can even give it a run here and try to test the step and see which
95:46 choose. If we go back we can see okay look it's used the chat model as the
95:51 brain and it's triggered the NA10 workflow as expected. You can see here
95:54 that it's sent off information to our other workflow. It sent the lead name,
95:59 the email, the message, and the company information, and it set it as qualified
96:02 as well. So, all of these fields have been filled out. We've got a nice AI
96:05 generated summary here from the model and brain. And we have the qualified set
96:09 to true. And so, the final step now for us is to head back over to our other
96:14 automation and just finish it off. Oh, we need to save that. I will
96:19 just run that again for you so you can see it in slow motion. It's using the
96:22 LLM as the brain and the tools agent and it's deciding whether it's qualified or
96:26 not. And if it is qualified, then it will send it to this workflow. Bam,
96:31 we've sent it. And there you go. If we this. Head back over to our qualified
96:40 lead classifier notifier. Now, we can add on a quick few steps here. I'm just
96:43 sort of going to rip through this. Um it's not super important. Um but it just
96:46 shows you a little bit more functionality of what you can build in
96:49 on N10. So, we're going to add in a messenger model step here, and we're
96:57 mini. And what I want this to do is to take in that information that we sent to
97:00 the workflow about the company research, etc. So, we know this is a qualified
97:03 lead now, but we just want to split it between either our SAS team or our
97:07 development agency sales team. So, they're specialized in dealing with
97:09 different cases. So, I'm going to cheat and just throw in a prompt here, which
97:13 you guys will be able to get access to. Um, which is basically saying we have a
97:17 new inbound lead. Um, change this to an expression. Sorry. Um, we have a new
97:20 inbound lead that we need you to classify into either SAS or development
97:24 agency. Here's the lead information. Um, we need to go back step and test
97:30 this. There we go. We should have some information. Um, and now we can put
97:34 these in. So, you see how there was nothing here before I went back and
97:36 tested the trigger so that it gives us some null values here that we can fill
97:41 out. Here's lead information. Um lead name uh message um name
97:51 request company information if the company is a SAS output SAS if the lead has development
97:57 agency upput agency. So we're looking for just agency or SAS as the outputs
98:04 here. Um simplify the output. Yep, that's all good. So we can no point in
98:07 us testing that step there because all the values are null. Um but the next
98:12 step is a basic router flow if so this is a basic conditional
98:19 routing. So we have the conditions we can go expression here. So we can go
98:24 um the content here. So this is the output from the open AI step. If the
98:29 content which is the response from the LLM step the classifier it's either
98:32 going to be agency or it's going to be SAS. So if it let's just to to make it a
98:36 bit more flexible. If we go string, if agency, great. So, if it contains agency
98:46 on the true side, we want to go Gmail and we want to send a message. Um,
98:54 and then if it is false, we want to do basically the same thing. Now, I've got
99:11 Um, okay. So, here we're not getting much data on the input side here and we
99:14 can't seem to simulate it because it's of course triggered by another workflow.
99:18 What we can do is just save it here. go into executions. And if we go back to our
99:29 agent, and we can go to the form submissions one. If we go to executions, and we just
99:36 run one of these that we just did before, copy to again. This is basically just going to
99:48 trigger this again. and so that we get a fresh execution and we can sort of pull
99:52 that data back into the workflow. Oh, again. Boom. Triggered it. And that's
100:28 all done. Now if we go back to this and we go to executions, go to the most
100:32 recent one that succeeded. It's going to load in. Oh, hang on. This one's
100:51 it. Okay, so this one here, if we click this, yep, we've got all the data in
100:55 here. So, what we can do is copy this into the editor and then we've got the
100:58 data that we need that's already loaded in so that we have some values to put
101:02 into our Gmail steps Gmail. So, that's a handy little trick to to know how to do.
101:07 And now we have all of this information. So, that's what that's what I was trying
101:11 to get. Um, the same setup and we're going to send this to I'm just going to
101:15 use an example here and call this um it's the same email. You wouldn't pull
101:18 this in necessarily. I'm just using this as something that I can show you
101:24 at show. Say new agency lead. Let's do a text. We say um new agency lead man. Go
101:30 get him. Um turn into an expression. And then we say we can just
101:37 throw this company data in there. It's going to be messy. You can play around
101:40 with this more when it comes to formatting, but just to show you the
101:44 functionality. If we go uh test a step here, that's going to sent an email to
101:49 this. This is like my agency sales reps uh email. Of this. Can duplicate this. Right click,
102:06 duplicate, bring it here. Oh, connect this up. And we change this.
102:14 You change this to your like SAS guy. You change it to a different different
102:18 email. Um, of course, and then you can say new SAS lead. Right. So now we have done all of
102:27 that basically all built out. The data is going to come in from the agent. It's
102:30 going to send in the company summary. This is going to classify it into being
102:34 a uh agency lead or a SAS lead because those are the only two types of
102:37 businesses that we work with. So all of them will be qualified when they come
102:40 through here. And then it just sends an email to our uh agency sales rep or our
102:46 uh SAS oh rename this to our SAS sales rep, new SAS lead for them to continue
102:56 with. Right. So to test this we can turn this on to active and you can see that
102:59 you can now make calls from your production form URL. Um we can go okay.
103:05 If we double click on this we can open this up. We can click on production URL.
103:11 Copy this and open this up in a new tab. spin. So of course my agency Morning
103:20 Side AI does development services. So this should be qualified and it should
103:24 also route it to the agency email. So if I now go submit, we go back into NADN, we go into
103:32 executions, we can see this one is running. If we go to inbox and there we go. We see it has
103:47 succeeded. And then if we go to and then if we go to our lead
103:51 qualifier and notifier and we go to executions, we will also see that we
103:55 have a new one that has succeeded here which was just uh a few seconds ago and
104:01 that's gone through. It is um outputed it as agency which is the the
104:05 classification that we wanted. It has gone through and has sent a new email.
104:08 And if we go to here we have new agency lead there. There we go. All the
104:11 information. So that's working number one. Now we can go back to our form and
104:14 we can try it again but this time with let's say an unqualified business. Let's
104:22 go. What is your name? Ray Croc Ray McTum I need more guys more people
104:33 flipping damn burgers. So essentially Ray here has come to our recruitment agency and
104:39 they're asking, "Hey, I need people to do flip patties for me in my fast food
104:43 restaurant." Um, and because Big Boy Recruits in Dallas, which is a
104:47 hypothetical company, of course, um, doesn't do that. It's going to qualify
104:50 them or it's going to disqualify them and then send an email to our good
104:54 buddy. Oh no, some poor dude at McDonald's is going to get an email now. Um, because
105:00 send an email. Um, but it's going to be running and of course it's going to be
105:04 sending. if they're unqualified, it will send an email to them and say, "Hey,
105:07 sorry you're not a good fit for us. Let us know if we can help or we can connect
105:13 partners." And while that is running, I would just put together the final one
105:16 here to test the functionality, which is if we go Liam admin.com, and we set up my SAS
105:24 https, my SAS agent, if you haven't already used it, we're going to show you
105:26 how to use it in the last tutorial of this video. So, you guys will get to see
105:30 that. Um, which is my own no code AI agent building platform. And what can we
105:34 help you with agents? Um, so this should be a SAS one and it's going to qualify
105:45 have the McDonald's one has succeeded here. And you see, yep, as expected, we
105:49 were not qualified. The McDonald's person was not qualified for our offer.
105:52 So, it looked at the qualification criteria we provided in here, said,
105:55 "Hey, no, that's not a good fit." So, I'm going to use this tool. And you can
106:00 see that it sent the email and it said, "Hey, thanks for your interest. Um, but
106:04 we're not a good fit for you." So, someone at McDonald's just got an email.
106:07 Apologize for that, but we didn't trigger the other workflow, which is a
106:10 key part. And we're not going to send emails to our sales team saying, "Hey,
106:14 look, new leads." Now, I have sent another one through here which just
106:17 finished executing. And we can see this. It's gone through. It's researched um
106:24 agentive. you'll see um Agent is a leading service delivery platform for AI
106:29 agent AI automation agency owners um etc and it's called the tool because we were
106:32 qualified because we're a SAS business right and again if we go back to
106:38 here and we look at the most recent one here then you'll see new SAS lead has
106:48 been triggered because we are a SAS of course um the LLM step here has outputed
106:53 just SAS So that means that it should send an email to the SAS team, which if
107:02 inbox, tada, new SAS lead, right? So I know that may have taken a while,
107:06 but uh we got there eventually. And you can see that we've built out all of this
107:09 functionality. We have our AI agent calling our tools if they are qualified
107:12 and triggering this other workflow. Again, you can build so much cool stuff
107:15 by connecting an agent to multiple different workflows. We have a little
107:18 relevance AI researcher tool that we're reusing here and we have people getting
107:22 denied um with an instant email sending them back. So, hope this been a cool one
107:26 to show you how NATM works. I really, really like this agent functionality
107:29 that they have. I think you guys are going to be able to build some awesome
107:31 stuff if you keep going down this rabbit hole. So, that has been agent build
107:35 number two. Stick with me as we jump into agent build number three, which is
107:39 a pretty damn cool one, focusing on both chat and voice-based agents all in the
107:43 same build. So, let's get the ball All right, so that is two builds out of
107:51 the way. Well done if you made it this far. We have another big one here. Um,
107:55 this is going to be breaking down how to use voice flow. Let's take this off
107:59 here. Um, to build an agent that is going to be both accessible through a
108:02 website chat, so you can chat to it on a website as a as a chat widget, which
108:05 you're probably familiar with, but we're going to connect the exact same agent
108:08 and exact same functionality that you get through that chat widget also to a
108:12 phone number on the website that will be able to call and have the same
108:14 experience. So, this is going to show you how on voice you can build both chat
108:19 and voice experiences. Um, and this is a new feature for them as of the time of
108:21 filming. And this agent is what we can classify as an AI customer support and
108:25 lead generation agent for both website and phone. And we're going to build it
108:28 on voice flow, of course. And the purpose of this agent is that it's
108:30 designed to be able to answer common questions from potential customers via a
108:33 website chatbot and also via a phone number that can be called. Not only can
108:37 it answer questions to help them sort of move them towards a uh a purchase, but
108:41 it can also generate instant quotes for interested parties. This is intended to
108:44 increase the number of leads that they get because people who see a contact
108:47 form may be like, "Hey, I want to get instant response. I want to know
108:50 instantly how much this is cuz I'm shopping around." Um, and rather than
108:53 just filling in a contact form and waiting. Um, having this instant
108:56 quotation can give people confirmation on the price. Um, so they're ready to
108:59 take a step forward and end up getting the sale ultimately. So that instant
109:02 quotation feature is a cool one. Um, very easy to do with the custom tool on
109:05 relevance that we're going to build. And finally, this agent is going to be able
109:08 to actually capture lead information from those who have been given a quote.
109:11 So after they've been given a quote, then we'll move to say, "Hey, give us
109:15 your details and we will follow up. Our team will follow you up and set an
109:17 appointment for the service." And the value here of the system is that
109:20 customers are going to often want instant answers so that they can make a
109:23 purchase. So by offering easy ways for them to get this information, we can
109:25 increase the chance that they're going to purchase from the business. Um,
109:28 companies typically have to spend money on some kind of customer support or
109:32 sales team in order to get these kind of answers given to customers when they
109:35 need them. But this agent can essentially be a oneanddone solution um
109:39 to both help increase the sales of the business by increasing that likelihood
109:42 of purchasing because they now have more information um while also saving the
109:45 business money that they would typically spend on some kind of support staff um
109:50 say if this chatbot can handle a dozens and dozens of responses a a week that
109:53 would typically have to have gone through a support person then we're
109:56 saving the company money and also helping them increase their chance of
109:59 generating more revenue. So um here's a rough layout of the design here. We are
110:02 going to have a website. I'll give you a template for this. It's very easy to set
110:05 up and we're just going to throw in a number um that's going to be connected
110:08 to the voice agent that we build and we're going to be setting up voice flows
110:11 web chat widget as well. And this agent is going to have access to a knowledge
110:15 base to answer questions um that prospects may have about the business
110:18 and their services etc. Um it's going to have a tool that is allowing them to
110:21 generate an instant quotation. So it's going to take in some information. This
110:24 is going to be for a cleaning business or a hypothetical cleaning business. And
110:26 then we're going to be able to generate an instant quote for them based on the
110:29 property type and the size of the property they need cleaned. and we're
110:32 going to be able to capture the lead information afterwards and log it into a
110:35 CRM. In this case, we'll just use Google Sheets, but it's fairly easy to swap
110:38 that out to whatever CRM you want. So, it's going to look a bit like this. I've
110:41 actually added a little bit more. And we are using another relevance tool in
110:44 here. This is from a different project from my accelerator, but I'm going to be
110:46 pinching that and putting it in here for you all. And this is the uh tool number
110:51 two here, the generate instant quote. So, we're going to be slotting that in
110:53 there, taking in some information, answering questions, etc. The process of
110:56 building on Voice Flow is one of my kind of favorite experiences um in the
110:59 automation space. I really like the the way they've built out their uh their
111:02 flow builder. Um so I'm sure you guys will enjoy building this uh step by step
111:05 with me. And then the general usage pattern of the system is that the
111:08 person's going to arrive on the website. They'll either click to chat with the
111:11 chatbot and engage with this functionality or they'll enter the phone
111:13 number into their phone. And then the agent will jump in and respond either
111:17 through text or through uh voice and determine what they're needing help
111:19 with, which is this section here. And then it's going to be routing using this
111:23 router section here to the correct tool whether they want a question answered or
111:27 they want to get a quote. Um, and then each of these branches will execute on
111:30 that uh, functionality depending on their intent. So, it's going to look a
111:33 bit like this. We'll have a phone number and we'll have a chatbot like this. This
111:36 is actually an agenda chatbot from my own software, but we'll be swapping this
111:39 out to a voice flow one in this build. So, without further ado, let's jump into
111:42 voice flow. So, when you click that link on the Figma, it's going to take you to the
111:47 signup page. You can sign up there and then once you're in, you're going to get
111:49 a page that looks a bit like this. The first thing that we want to do, of
111:52 course, is to create a new agent up here on the top right. Let's call this
111:56 Bonor's cleaning um, website. and phone agent. Um, let's just start with a basic
112:00 template here. Import knowledge for this import knowledge. We can actually just
112:03 skip that for now. And then we get into the flow builder on voice flow. So just
112:07 a quick orientation if you are new to the voice flow platform. This is where
112:10 we can add in our knowledge which we will do shortly. The workflows are where
112:13 we access the flow builder. In most simple builds like this, you just going
112:16 to have one workflow. So you don't need to worry too much about that. Now we
112:20 have integrations like uh the widget which we're going to be using to deploy
112:23 this on a website. We have the phone number integration which we're going to
112:26 be doing later as well. Then we have API keys etc which you don't need to worry
112:29 too much about right now. We have some publishing features here which we'll
112:32 double later. We have access to transcripts. So once we deploy this you
112:35 can access all of the transcripts either by voice or through chat here and and
112:39 sort of dig through the answers and and see how the uh people are interacting
112:42 with the agent that you've built. Um something that a lot of people neglect
112:44 after they've put one of these into production. And then we have things like
112:48 analytics um etc. But obviously we need some data before we see anything useful
112:50 there. And then the settings page is not too much you need to worry about right
112:53 now. Just sort of on a need to know basis. The more important stuff, of
112:55 course, is up in this first tab here, which is content. So, we have messages,
112:59 we have prompts, we have components. We're going to be working a lot with
113:02 prompts shortly. So, that's the main one we need to be worried about. But for
113:05 now, we can just go into workflows, and we open up this first workflow and edit
113:09 it. And here we have the template that VoiceFlow gives for us. Um, which I'm
113:12 actually just going to nuke this, and we'll start fresh. And if you see on the
113:15 Figma, we have a design here that we're roughly working towards. I'm going to be
113:18 showing you the sort of step by step. So, we need a welcome step.
113:22 So, we're going to start off by going here and dropping. I'll try to zoom in a
113:27 bit for you guys here. Talk message. So, message is how you send a message um via
113:31 the chat. So, start is when maybe you click on the widget, it pops up. And
113:34 this message that we're about to put in is the first message the bot is going to
113:39 send. So, we say um up here, hey, hey, welcome to corners. I'm going to zoom this up for a
113:44 bit for you guys. And right away, we have a little tip and trick that I want
113:47 to give to you because we are building this as a chat and voice assistant. We
113:51 want to over voice. You don't want to overuse punctuation because it leaves
113:54 these big long pauses when the the voice agent is going, "Hey, welcome to
113:59 Connor's cleaning." So, we wanted to just say, "Hey, welcome to Connor's
114:01 cleaning." A bit more natural. There's times where you'll see me on the side of
114:04 sloppier punctuation, but that's just to ensure that when we get to testing it on
114:08 voice, it sounds as natural as possible. So, hey, welcome to Connor's cleaning.
114:10 And then, of course, we're going to wait for them to reply and say something back
114:14 to us. We go to listen and then capture. And this is going to capture the
114:17 information. We want to change this from capturing entities which is like say I'm
114:21 looking for a price or an address. Um we're just going to go the entire user
114:24 reply and the reply is going to be saved into this variable here. So we actually
114:30 want to change this to um first user reply because we're going to need it a
114:37 little bit later. Um the users first reply. And I like to name these as we
114:41 go. So we can call this welcome. Drag this out here. Get a new
114:45 one. And we're going to be doing a uh a set step here. So, we're going to be
114:48 setting some variables. And I'm going to add a new variable to set. And we're
114:52 going to do it based off a prompt here, which is a cool feature in voice they've
114:55 added recently. And we're going to be able to select a prompt that's going to
114:58 take this information from this first reply and then generate some kind of
115:01 output from it and set a variable. And the variable we're going to set here is
115:04 called last response. So, this is typically what you're going to put as
115:06 the last response from the AI or from the agent. Um, last response here. And
115:10 last utterance is typically the most recent uh message from the user. So
115:13 utterance is coming from the user the most recent last utterance and the last
115:18 response is what the the AI or the agent or the system has last responded to. So
115:23 we want to set the last response to something that is generated through this
115:25 prompt here. So we can create a new prompt here and this is basically going
115:28 to take in the data from the chat and the conversation so far and we'll be
115:31 able to generate things off of it. So say we add in here the conversation
115:35 history. That's a good thing to have in in most cases. And I'm going to be
115:37 dumping in some of the prompts here just to save us a bit of time. But basically,
115:40 we're saying summarize the customer's question below and ask them to confirm
115:43 that that's what they meant. And so, we're not actually going to be
115:45 generating the last utterance here. We're going to be adding in the last the
115:50 first user reply that we got. I mean, it's going to be included here in the
115:53 conversation history, but there's no harm really in hard coding it or at
115:56 least putting the variable in here to make sure that it's in there. Um, we're
115:59 just saying summarize the customer's question and basically say a
116:01 confirmation statement. So, just to confirm this is what you're looking to
116:04 do. So, you imagine this over the phone. Hey, um um yeah, I'm not really sure
116:08 what I'm supposed to be doing here, but I was thinking if if you guys were
116:12 possibly so all of that information is taken in, we can flick back to them and
116:15 say, just to confirm this, this sounds like you're looking for this, yes or no.
116:18 Um so, ensure your tone is empathetic. Speak directly to the end customer. Keep
116:22 your answer brief and two sentences max. So, if we go back here and actually we
116:28 can name that prompt. So it's um summarize problem and then we need to send the
116:35 response. So this prompt is going to take in the information we provided
116:38 here. It's going to use this prompt to take in um the conversation history so
116:43 far and this information from the user in the first question. It's going to
116:47 generate a a question to ask back and it's going to save it to this variable
116:51 that we have here. So apply output to variable last response. There is
116:54 actually an easier way to do AI responses like this, but in our case, we
116:57 need to be saving this variable. So, it'll make sense in a second. But, we
117:01 can go into here and we can go last response. And then it's going to send
117:05 that information back to them. So, let's just do a quick test here. Click start.
117:11 Hey, welcome to Connor's Cleaning. Oh, actually, we need to ask a question.
117:15 Hey, welcome to Connor's Cleaning. Um, We'll say I need cleaning services for my
117:30 house. Sounds like you're looking for cleaning services for your greenhouse.
117:33 Is that correct? Want to make sure I understand your specific needs before we
117:36 proceed. So, I obviously spelled house with G house. So, we thought it was a
117:39 greenhouse, but that's what we want. Some kind of confirmation message just
117:41 saying like, hey, look, is this what you're actually looking to do before we
117:44 then go and trigger the different tools that we are equipping our agent with? By
117:47 the way, there is a way of changing between trackpad and mouse. So I am
117:50 panning around with my mouse here. You can also do a trackpad method which is a
117:53 lot easier to use if you're if you're having trouble using it. Okay. So after
117:56 that they're obviously going to say yes or no whether like have I got the
117:59 question or have I summarized what you're looking for correctly and we can
118:03 go to a choice step here and we can set up some triggers. We can set the intent
118:07 to yes and then we can add another trigger and we can set it as no. So this
118:12 is basically using AI to analyze what they've said and grouping it around
118:15 these certain things which are called intents. So, what is the intent of them
118:19 of this uh of the response? And in this case, they have some pre-built ones, but
118:21 we are going to be building our own custom intents later. But for now, just
118:25 know that if you're looking to sort of split traffic or split people coming
118:28 through the system, these choice blocks with the default uh intents from voice
118:32 flow, yes and no, are ready to use out of the box. And if they don't say yes or
118:35 no clearly or we can't pick it up, we can add a no match here. We can say
118:39 sorry, I didn't get that. Can you say yes again? A yes or a no is enough. And
118:45 then we can say to follow a path after these reprompts which we'll call no
118:49 match. And then we have this uh no match path which we'll set up in a second
118:53 here. I'll just put it as a a filler for now. Basically if if they don't say yes
118:56 or no um this is setting up error handling. Um, and basically if people
119:01 particularly over the phone, um, there's so many different ways that the
119:03 conversation can go and end up and you'll want to, while I don't focus on
119:07 it too much in this build, um, as you're building production grade assistants,
119:10 you'll need to build a lot more of these fallbacks and these reprompt and these
119:14 no match things to handle edge cases where people use it in a weird way that
119:17 you don't expect. So, I want to give you a little taste of that in this tutorial,
119:20 but it is nowhere near representative of what it takes to actually get something
119:23 into production that you can trust on a on a customer's website. Okay, so we've
119:26 got this choice block set up to determine if we have got their
119:29 summarization of the problem correct. And we can take this up to here and we
119:33 can set another uh variable. And so this is really the core part of the
119:36 application which is determining what their intent is, what are they looking
119:39 to do and which tool are we going to route them to it. So this is a very
119:41 handy skill to have which I'm going to teach you which is how to set up some
119:46 kind of intent classification system. Um which is really really essential to
119:49 building agents on on voice flow and any kind of agent where the platform itself
119:52 isn't automatically handling that for you. So if we go set a new variable and
119:56 we're going to do it through a prompt. We're going to set a new variable here
120:00 called desired action. So basically people coming through and asking questions can
120:07 be saying hey look I I just have a quick question about where you guys are based
120:09 and then we're going to route them to the knowledge base. And then someone may
120:11 be saying hey how much does it cost for this? And then we're going to route them
120:15 to the pricing uh the instant quotation um system that we're setting up. So
120:17 needs to be able to determine what they're looking for and we're going to
120:20 route them depending on that. And that's what this router is going to do. So the
120:25 action that the prospect wants to take, the most likely action that the prospect
120:29 wants to take. Then we need to make a new prompt here. We're going to call
120:34 this intent classifier. Classifies the intent of the uh prospect into asking the knowledge
120:44 base or generating an instant quote. Add in the conversation history.
120:47 It's always good to have that in there. And I'm going to put in the prompt that
120:49 I've written previously. And this is a pretty basic one as well, which is just
120:52 saying what does the customer want to do, ask a question, get a real-time
120:55 quote, or something else entirely. You must output a label for this only. Your
121:00 options are ask a question, get a quote, or other. And you guys can just pause it
121:02 and see what I've got in there. But basically, anything asking a question is
121:05 going to be about the business and the services. And if there's anything about
121:09 pricing or directly related to getting a quote and like they're ready to move on
121:12 this, then we're going to route them to get a quote. And anything else is going
121:14 to go to other. And because in the next step, we are going to be looking out for
121:18 either this is the output or this or this. a really clear statement saying
121:21 this is all you need to output just this and not hey I took a look at the the
121:25 conversation history and it seems like the user wants to do get a quote we just
121:28 want just get a quote and we can explicitly state that with this big caps
121:31 lock block here and as with the other builds all of these prompts are going to
121:34 be available in the resources for you to follow along with okay so now we get to
121:39 the cool bit which is routing this. So if we go condition add this in here and
121:46 we go add path condition builder and we say if desired action
121:51 is ask question. Oh that's that's all we need there. So as you can see that's added
122:02 one in here. And if we want to add another one in, we go if desired action
122:11 is um what do we have the label as output? What did we what's the exact
122:16 label that we had in here? So this prompt is going to be outputting these
122:19 labels. So we need to make sure they match up. So ask a question. Get a
122:29 Yep. Get a quote. other and it's already got an else path in there for any error handling as well.
122:39 So what this now allows us to do is to build out our different tools. So we can
122:43 go up to here to ask a question. Just throw this in for now so we can get an
122:47 idea of what it's going to look Um other this is going to be sort of
122:54 error handling. And if it's else, that means that the LLM step here hasn't outputed
123:00 any of the labels that we told it to, and it's likely thrown in a bunch of
123:04 rubbish. Um, so this is sort of an error handling step. Um, should say, "Sorry,
123:09 something went wrong, at least during this prototyping phase. So now what I'd
123:11 like to do is make this look a little bit prettier. Um, we can go through and
123:16 add things in here like this is the uh confirm problem um intent classifier
123:23 router." And then we can go here add a note can say tool number number one
123:37 base two you guys don't have this so be easy for you guys to see understand I
123:44 confusing and then this other one we don't need to worry about too much so to
123:46 keep things quick I'm not going to test this just yet I'm going to test it once
123:49 we've got that functionality set up on either side at least for this top one
123:52 first so now We need to go and set up our knowledge base. And to do this, we
123:55 can click on the back button here. Go to our knowledge. And here we have a data
124:00 source which we can upload. I'm going to upload a file, but you can put in URLs
124:04 to different websites, etc. I'm going to be uploading a file
124:08 here. And I'm going to upload this Connor's cleaning FAQ kind of document,
124:11 which you guys are going to have access to in the resources. Basically, it's
124:15 just about us, location, our services, ra um some frequently asked questions,
124:20 etc. So, I've just AI generated this. Um, and if you're doing any kind of
124:23 prototype builds, I recommend you do the same just to throw it in there and see
124:25 if the knowledge base is working as expected. Obviously, you'd swap this out
124:28 with actual customer data or or your client's FAQ. I'm just going to throw
124:32 that in there for now. And you guys can do the same. And when you're setting up your
124:38 knowledge base, you can also set up the settings for it. So, in this case, it's
124:42 using by default Claude 3.5 Haiku. And you can see how many tokens this is
124:45 going to cost you for Voice Flow's usage. Um, what is Haiku? Haiku seems to
124:49 be the cheapest. Oh, you've got GBT40 mini. Let me just chuck on GBT40 mini
124:53 here. We want this thing to be pretty deterministic. So, I'd say 0.1 is fine.
124:57 Max tokens. Um, we can increase that just in case it needs to give a longer
125:02 answer. And chunk limit of of three should be enough. So, that's just so
125:04 this stuff is a little bit more advanced. That's that vector database as
125:07 I was talking about. Basically, knowledge base is going to be sending
125:11 the message that we ask it and querying it and getting back chunks of
125:14 information. Because our knowledge base is quite small, we don't really need to
125:16 have too many chunks. If you put this up, you just be getting the whole
125:19 document back. Anyway, so max tokens, the number of tokens that it's going to
125:22 include in the response. So, we want to increase that to 480 um so that it can
125:25 give a longer response if they need. Maybe just tone that down a bit. And
125:28 these are of course kind of controls that you have on how much you want to
125:31 allow the app to spend. And those settings obviously the main ways that
125:34 you can control how much um your knowledge base is using and how much
125:37 your you or your client are ultimately spending on the AI features for the
125:40 knowledge base. So once we've got that set up, we can go back to workflows
125:43 here, open this up again, and then to plug in our knowledge base, we can go to the dev section here.
125:49 We can go to KB search, pop this in here. I will uh we need to delete that and reconnect
125:57 this up to the top. And we're going to delete this as well. And so we can go
126:02 into this knowledgebased step here, and we can enter the query. So what we're
126:05 going to say is we basically want to throw the information that we've got
126:08 from the user already about what they want which is we have here as the first
126:12 message they gave us which might be a bit longwinded. Then we have the summary
126:15 that they have confirmed and then we can put these into the query that's going to
126:19 be asking the knowledge base hey this is what we want information on can you give
126:24 us some information back. So we can go user first message put a curable in
126:29 there. We can go first use reply and then we can also go summarized problem.
126:33 Now you can see why if we put last response here, why we have this variable
126:36 saved instead of just sending it automatically. So I actually don't like
126:39 using last response because that's something that you like to update quite
126:41 a lot. So I'm actually just going to switch this to um changing it to
126:49 summarized problem just so we don't get any kind of overlaps that cause problems
126:55 down the line. A summary of the user's Then we put this in here. Get to spit it
127:07 out. So when you put a variable in a message, it's just going to print out
127:10 and and spit out the the value that's inside that variable. So we've set the
127:13 summarized problem variable and then we're just going to spit it out and send
127:16 it into the chat or over the over the phone. So now we can come back out to
127:19 our knowledge base and we can take out the SL response and replace it with
127:23 summarized summarized problem. Then we can save the chunks that come back from
127:26 the knowledge base. I don't want to get too in depth on what chunks are
127:29 specifically. It's a little bit more advanced, but for now, we can just know
127:31 that it's going to return some information from that knowledge base.
127:34 It's going to chop up that document we put in. And when we put in this
127:38 question, it's going to basically ask that knowledge base, can we get three
127:41 chunks that most closely match the information that is in this query that
127:44 we sent to it. So, we can save these chunks, which is going to be three
127:46 because we set that up in the knowledge base settings into this chunks variable.
127:51 And the chunk limit is three still. If we click this, we can add in a chunks
127:53 not found path. But for this tutorial, we don't need to worry about that
127:56 necessarily. And then we're going to use those chunks that came back from the
127:58 knowledge base to generate an answer based on the original question. So if we
128:02 put this here, we go talk, we go prompt. And for this prompt, if we go here, we
128:06 can create a new prompt. We can add in the conversation history just for good
128:09 measure here to give it the full context of what's going on. And then I have a
128:12 prompt here. You are an AI customer support rep from Connor's Cleaning
128:15 helping customer with the question. Use the provider details below to answer the
128:18 customer's question. Ensure you keep your answer brief and speak directly to
128:21 your end customer. You are speaking to them over the phone. It's the input data
128:25 provider details which is the chunks variable. I'll just put that in
128:30 again. Chunks to make sure it's set up properly to the user's original uh
128:36 question. We can put all this information back in. So, first reply,
128:39 this is what they asked us as soon as they picked up the phone or the first
128:42 message they sent when we asked them what can we help you with. And then we
128:46 also just put in for good measure our summary of the problem that they
128:51 confirmed. And we can go to our summarize problem variable and throw that in. That should
128:56 be good to go. And I like to make these look purple or some kind of cool color. Um, call
129:04 this a KB query. And we can change this to generate answer from chunks. Let's say
129:13 from Kh. And if you want to make this a bit easier for you to kind of understand
129:17 at a glance, you can add in your descriptions on these. So if we go to
129:22 edit again and we go here, this takes in chunks from the from the KB and their
129:28 original question and writes a short and sweet answer. And we have this in here
129:31 that you are speaking to them over the phone because we want to make sure that
129:33 we're building this with the phone in mind, which is more tricky than just
129:37 chat. So long text outputs don't really work that well over the phone. So that's
129:39 kind of why we're putting that in there as well. And we can call this um
129:44 generate answer. All right. So now we can actually give this a spin. We can start
129:55 again. Oh, we may need to if we just click run. Okay, there's no training
129:59 needed yet. We can run test. Where are you located? Sounds like you're asking about
130:05 a business location. Could you confirm if you'd like to know the specific
130:08 address or where Connor's Cleaning operates? Okay, it's good this popped up
130:10 because as you can see, it's asking for a non- yes or no answer. We're just
130:15 looking for a confirmation in yes or no. And so this would technically break the
130:18 system and that'd be saying, "Oh, I'd like a specific address." And this is
130:21 looking for yes or no. And so it would send it to this no match. So what we can
130:24 do to fix this is to go into the summarize problem prompt, modify the
130:29 prompt, and then say they should be able to answer only with yes or no. This is a
130:35 confirmation step, not asking for more located? Sounds like you want to know
130:56 the specific location of our cleaning business. Is that correct?
131:02 Yep. Now it goes to the router here. It's going to determine that I said yes.
131:06 Bam, bam, bam. And there we go. So that that all happened pretty quickly, but
131:09 you can see it's sort of broken down through here. Um, if I click over here,
131:13 it's going to remove all of this. Okay, so let's break down how this happened
131:16 step by step. Um, so you can see this through here. It's still using 3.5 haiku
131:20 for some reason. I'll need to double check why that's still using the model
131:24 we didn't select. But basically, it comes through this step here is the
131:27 intent classifier. So you can see that it set it as yes. That is the correct
131:31 intent and predicted intent yes. And that routed it to this. And then using the model again, it
131:38 analyzed the information that it was given. And then it set the desired
131:42 action variable to ask a question which is one of the labels that we wanted and
131:45 that is correct. And then it said condition matched taken path one. So it
131:49 set the desired action variable to ask a question. We were checking for it here.
131:52 Then it said okay great. Now I'm sending it up here to the knowledgebased query.
131:57 It says it's query received. We passed in all the information whereabouts you
132:00 located and then the summary that we gave it. And then we got two chunks back
132:04 from the KB and the AI response here finally took in all of this information
132:08 and it gave us the final output and generated this response. We are located
132:12 at 247 ra and at the end here it's saying is there a specific area you're
132:16 interested in. For a basic build like this I'd probably change the prompt to
132:19 say don't ask another question because in this case you then need to set up a
132:22 looping mechanism where it can keep answering questions for them and then
132:25 break out into any of these other intents um as needed. But for now that
132:29 is a knowledge base and that's how you can ask questions. And so that is tool
132:32 number one knocked out which was easy enough. So great we can go on to tool
132:35 number two now. So for this second quote I'll be able to give you a relevance
132:38 tool that we're using for it. But let's just jump into answering this question.
132:41 So they've said here that they wanted something related to pricing or quoting.
132:45 Remember in here in the router we have set the intent classifier to say it's
132:48 going to go to the get a quote if they're asking about pricing or have
132:52 directly requests a quote. And this will take them to a real-time quotation tool
132:55 that takes the property type and size and then returns an estimate. So that's
132:58 the people that are going to be getting to this next branch of the agent. So
133:04 desired action is get a quote. We can say, "Okay, sure." To give you an
133:14 just Okay, sure. To give you an instant quote, I just need the properties type
133:18 and size and square feet. Then I'm going to add another chat step or message,
133:32 apartment? The next we can do one of these choice steps again and this time
133:36 we get to create some custom intents. So we can go to triggers here and we want
133:39 to select an intent. See it doesn't have a house. So we can go create an intent
133:51 users property type is a house. And then we can add in some examples here. So
133:54 obviously house this is just giving examples to voice flows AI engine to
133:58 help us better to classify the different intents as they come into this step. So
134:02 we can say home and then you can add in some AI generated ones here which
134:06 usually pretty easy and uh which usually pretty good. Okay, residence, dwelling,
134:10 property, abode, living place. Uh this this gets tricky um because some of
134:15 these could overlap with apartment. So property is probably abodess too broad
134:18 living place. uh dwelling uh residence potentially we could get away
134:26 with. We could say like home single family home and by now we've given enough
134:39 examples where we can just go to create and now we need to do the same for
134:43 apartment. So, we'll go to create an intent apartment. The user's property
134:49 type. The user's property type is an apartment. Um, see what else it's got
135:02 for us. Um, townhouse, penthouse, duplex, flat, probably not
135:08 right. Loft probably not right. Townhouse penthouse. Think that's a good
135:12 bunch. And then we can add in a no match here as well. Um, and we can add in a reprompt. Sorry,
135:29 an If it is a house, we can come up here set. We can go value or expression. Select the
135:42 variable to set. We're going to go property the save that variable or create that
135:53 variable, sorry. And then we can enter this in and set this as
136:00 house. So we're setting the property type variable to house when it's been
136:05 triggered um by this particular route. And then we need to add another
136:11 one. So, I'll probably just duplicate this. That's by right clicking on one of
136:15 these blocks. Um, and then we can connect up apartment to it. Property type. Instead of being
136:25 apartment. And from here, now that we know the property type, we can set the
136:29 size of the apartment or the house. So, we go and how many square and how many square feet is it?
136:39 We can connect both of these up here. And then we're going to save the entire
136:43 user reply or on a capture step. So anything that they respond to after
136:52 thing. And we want to go here, set up a set using a prompt. And what we're going to
137:10 do is use AI to analyze the response and then extract the number of square feet
137:14 from it. We could have used a step here where we change this to instead of the
137:17 entire user reply, we change it to an entity. Um, but it's not as reliable as
137:21 doing it this way. So, I'd prefer to just get it right the first time. Um,
137:26 cuz a entity of the number of uh number of feet may be a bit harder to pick up
137:30 than a clear word like an entity of house or apartment or a name, etc. So
137:35 just to make sure that it works every time for us, we want to create a new
137:39 prompt. And I'm going to put this in here. Extract the number of square feet
137:42 from what the user said in numerical format only. Each 500 include nothing
137:45 else in your response. This will be saved as a variable and passed to an API
137:49 to a quote generator function. So giving it a bit of context about what's been
137:52 going on. And also put in the memory there just for good measure. Oh, we didn't name the prompt.
138:05 And then we need to set the variable that we want to save this to, which is
138:08 going to be proper property size. So now we're saving that user's, the size of the user's property
138:19 in square feet. And then we need to do one kind of tricky step, but it's just something
138:25 that you guys will pick up as you go and uh as you build more of these. But the
138:28 next step after this is going to be sending that information to a relevance
138:31 tool. And the relevance tool is expecting not what's called a string,
138:36 which is just a number of letters or just text. Um, you can have a number as
138:39 a text, which is confusing, but basically it's just the format uh in
138:42 which it's being received in. So, because we're sending it over an API, we
138:46 need to be specific about the format. So I can't take this is essentially going
138:50 to be saving that 500 or say if we say it's a 500 foot property. This is going
138:55 to be plucking that out and giving us 500 as a string. In order to get the
138:58 response we want from relevance we need to convert it to a number and then send
139:03 it. So a little block of uh custom code here. Um know this was a no code
139:07 tutorial but I hope you can forgive me this. And all we need to do is put this
139:13 in here. So property size this is the variable. We're going to go property
139:19 size and we're just casting this variable as a number and then
139:23 reassigning it to the variable that it was before. So we're taking whatever
139:25 came out of this, we're saying, okay, can you just make it a number um and
139:29 we'll save it and sort of overwrite the existing variable. So now we have the number 500
139:34 in that case that we're ready to send to the next step. And then if that's gone
139:49 And then we have this JavaScript fail route here which you can just sort of
139:53 throw down there for now. Um there is a chance that this prompt outputs not just
139:57 a number. As you can see we're asking it for just this and therefore the property
140:04 size variable if it just is 500. Um but sometimes it can say hey your size is
140:11 500. and then we end up with a variable that's not actually convertible into a
140:15 number. Um, so we do need to add a little bit there as a as a potential
140:19 fallback. Um, probably won't be doing it in this video if I'm honest. It's fairly
140:22 basic. Um, but you would add this in here some sort of looping back in to
140:25 make sure that it is actually in the right format or just make sure that your
140:28 prompt is actually only you put like a strict instruction only output just the
140:32 number um so that you get less errors there. So the next major step after this
140:36 is to get our relevant AI research. If we go back to the Figma here, you'll see
140:40 that we have this relevant AI tool which is going to allow us to uh generate the
140:44 instant quote. So, I've pre-made this and I've actually sniped it from some of
140:47 my accelerator resources. So, if you open this link, it's going to allow you
140:51 to clone this into your relevance account. So, up here you can click clone
141:01 AI. And now this thing is going to be taking in a property type and a size and
141:04 square feet. And then we have a basic LLM step here. This is a really really
141:07 simple one. Again, like I said earlier in this video. The building really
141:12 powerful and advanced functionality does take a lot longer. And if I was to do
141:15 things that weren't just a very basic LLM step like this, then it would take a
141:18 lot longer with different platforms you have to sign up for. The idea is to
141:22 teach you how these things can connect. And then you can come into relevance
141:24 here. Once you know how to connect voice flow to relevance through an API call,
141:27 you can come in here and throw whatever the hell you want in and make it as
141:30 advanced as you want. But in this case, we literally could have done this in
141:34 voice flow if I'm being fully honest. But the idea of being able to access an
141:38 external tool via an API is really the skill that I'm trying to teach you here.
141:41 So, what this is doing is taking in the property type. We've got two options
141:46 here and the square footage in a number. See here, it's it's a number. I can't
141:49 just type in here. It has to be a number. So, that's what the API is going
141:51 to be expecting. You can see here that it says it's expecting a number. And
141:55 then we pass this into a basic prompt here that's saying the customer is
141:58 requesting this um and these are the rough prices for the different square
142:02 footage and different types. and the LLM in this case GPT4 mini just going to
142:05 look over that take in their inputs and then give an output. So if I just go up
142:10 here and I set it's an apartment and I mean I don't know what five how big is
142:15 an apartment in square feet I don't know that it's going to give us some kind of
142:24 output and then it gives us the estimate. So it's saying regular
142:26 maintenance cleaning for 60 deep cleaning for 120 move and move out etc.
142:30 So that's the output that we're going to be sending back to voice flow and we're
142:33 going to be turning that into a nice message to send through chat or through
142:36 the phone. So in order to set this tool up in voice flow and be able to interact
142:40 with it, we need to we can hide this make sure the tool's been saved and you
142:43 have it in your account and then you can go to use here and just like we learned
142:48 in theuh NA10 tutorial we can go to API here and then we get an API for us to
142:53 use and again is a post request and it tells us how to use it. So we have
142:56 params basically the inputs it's expecting of the property type and a
142:59 string. You see how it's got two uh quotations? That means anything inside
143:02 it is a string. And the square footage here, you can see it isn't in quotation.
143:05 So, it's not a string. It's in this case, it's a number. And of course, we
143:08 have the project ID here, letting relevance know which tool that we've
143:11 created that we're actually trying to interact with. And so, all we need to do
143:14 now to interact with this and set up our quote generator is to copy this link.
143:17 So, this is the endpoint URL that we're going to be calling over the internet.
143:21 And this is what it's expecting in the in the body of that post request. If you
143:25 go back to voice flow here and we need to go to dev API, we're going to change this to post.
143:31 Put in that URL. And in this case, seems that we've got a little bit of a different page on
143:36 relevant for some reason. So we can actually do the authentication step,
143:38 which is helpful. So you can click generate API key here. May have changed
143:41 by the time you're watching this video, but should roughly be the same. And we
143:46 can click deploy here. Make sure that the API is up and running for us to
143:48 interact with. Actually, now that we've deployed it, we don't need the authentication
143:55 anymore. That was a bit strange. Uh, actually, I will do the API key just so
143:58 you guys see how this works. We can make it private here. So, now that we have
144:02 our API key, we need to see how it's expecting to receive that API key via
144:05 the HTTP request, which we're going to be sending from voice flow. Um, so if we
144:09 scroll down here, curl is usually the one I like to go to. And there is a
144:12 little bit I haven't really explained in terms of headers and bodies when it
144:15 comes to API calls. But for now, just know that when you have a curl request
144:19 like this, maybe it's even easier on the uh JavaScript. Here we have what's
144:22 called headers. And this is basically like the the the envelope that you put
144:27 the information in. So this up here is the information that we're putting in
144:29 the request. It's going inside the envelope. And the headers and the method
144:34 and the endpoint URL are like the stamp and the information that you put on the
144:36 outside of the envelope to make sure it gets where it wants to go. So you can
144:40 say that this endpoint here, that's what we call the endpoint. Same as up here.
144:43 That's the the same as we have just here. Endpoint is like the address where
144:47 the envelope is being sent to. We have the post, the method. Maybe this is like
144:52 super fast mail or like overnight delivery or maybe like a parcel versus
144:56 an envelope. Basically, the type of delivery that you're doing or type of
144:59 request. And the headers include important stuff like the type of content
145:02 that's inside it. So you might say this like this is there's written text or
145:05 there's a letter in here. So it might seem a little bit complicated, but we'll
145:08 fill this all in. Now, anytime you are making an API call, you'll see these
145:11 headers around and you'll see the endpoint and the method and then the
145:14 body which we're going to be setting up. So, this will all make sense in 2
145:17 seconds, but let's just say for now we have content type and it's going to be
145:20 application JSON. That's one of our headers. So, if we go back to Connor's
145:25 cleaning, we open up the headers. We can go content type here as we saw here,
145:29 right? So content type, we need to set it to application/json. Very common one that
145:33 you're going to be using. And we have another one which is authorization. So this is basically the
145:39 majority of the endpoints you're going to be seeing which is authorization and
145:43 content type. And a lot of the time it's going to be an API key and application
145:47 JSON. So then we can go back to relevance. We can get our API
145:54 key. And now to set up the body of the request which is this. We can copy this
146:01 and change this to raw. And we'll paste this in. we have the params or the
146:05 inputs in this case, the property type and square footage for the tool. And
146:08 what we want to do is insert the variables that we've got over here into
146:12 these. So we can go inside these quotations here and we can go property
146:20 uh type in this case and then for square footage we can go property
146:27 size. And so because the property type is expecting a string as we can see if
146:32 we go back here um to build It's expecting a string and this is
146:37 expecting a number. The string must be wrapped around in uh these quotations
146:42 and the square footage is a number and so we can just put the number because
146:45 we've already converted it into a number here. Right? So that's probably the
146:48 trickiest part of all the stuff I'm going to teach you today. And now that
146:50 we have this set up, we can actually send a test request and we can say uh
146:58 house. We can say 500 and test this API call. And there we go. It's complete. We
147:02 have got back our answer the estimate etc. and all of the information is
147:06 coming back from relevant AI as expected and now what we need to do is extract
147:09 the information which we want which is this answer from this API call and from
147:13 the response. So we sent a request and we got a response. Remember how we were
147:17 using those terms before. We can click on answer and we can save it to and we
147:22 can call it uh raw quote data because it is quite raw from. And now we have the information
147:37 back from relevance AI. All we need to do is make it pretty and use AI to
147:41 generate a message that summarizes the quote to the customer. So we can go a uh
147:48 prompt step here. We can create a new prompt. Let's add in the conversation
147:53 history for context. And I've got this prompt here. Write a short and clear
147:58 explanation of this quote for the customers. uh we can go property property
148:06 type for the customer's property type and we can put in the raw quote data
148:10 here. So that's going to insert this with the customer's apartment or the
148:13 customer's house and then we're just going to dump in the raw data that came
148:16 back from relevance AI. Your response will be read over the phone. So it must
148:19 be all in one paragraph and no longer than three to four short sentences. It
148:22 should read like and I've given an example here of how I wanted to give the
148:32 And so we can change this to quote quote response. And now we are
148:38 ready to give this a spin. So if we run this whole thing from the top
148:50 run, how much for a weekly clean? So I'm going straight to the asking a question
148:53 about price. You're asking about a pricing for our weekly cleaning service.
148:57 Is that correct? Yes. Bam. Bam. Okay, sure. To give you an instant quote, I just need the property
149:03 type and size and square feet. Is the house. How many square feet is it? 500.
149:15 I don't know. Is that a big That sounds very small to me. I don't know if 500 I have I have no
149:20 I have no clue about the sizes of houses if 500 is normal or
149:26 not. And there we go. Based on your 500 foot house, we have four cleaning
149:29 packages. Regular maintenance cleaning at $90 for standard weekly cleaning,
149:33 deep cleaning, ra. So, we're giving them a quick summary based on their 500 ft
149:38 house or x number of square ft x property type and giving them a quick
149:42 summary. So, that's pretty cool. We're using relevance tool and we're getting
149:44 this information back. Now, there is one more step that I have added onto this
149:48 this I get this thing where if I've gone this far with you guys, I may as well
149:51 add in like the rest of it to make it actually a bit more useful. um which is
149:55 a quick lead capture using Google Sheets and Make.com. So, I couldn't leave you
149:59 guys hanging on this. I thought I may as well throw it in there. So, stick with
150:01 me because this is really where you're going to be like, "Oh, this is this is
150:04 uh opening my eyes to to what you can do with these kind of platforms." So, the
150:07 reason we're adding this on is because the person has asked about pricing or
150:10 they're directly interested in some kind of services and we've given them an
150:13 instant quote and now we're trying to immediately follow that up with, hey,
150:16 look, give us your details and we'll be in touch and we'll get that service
150:20 booked in right away. So, we can jump into a uh message block here. This is
150:24 I'll just paste this in to keep things nice and quick here. Please provide your
150:27 name and phone number and I'll get one of the team to call you to find a time
150:29 that works. Now, one thing I will change is as you saw on that last run, it had a
150:33 question at the bottom. It's like, which one are you interested in? I would
150:38 probably try to remove that. Um, do not ask a question at the end.
150:53 So that's just going to end the message and saying, "Look, this is a quote."
150:56 Bam. And then the next message I get is this. Please provide your name and phone
150:59 number and I'll get one of the team to call you to find a time that works. And
151:02 so we want to save this entire user reply with a capture step. Capture step.
151:08 There we go. Change this to entire user reply. So we want to capture all of the
151:11 information that they send or over the phone or either through chat. So in this
151:13 response, they're going to say their name and their phone number, right? and
151:17 we need to extract those out. I'm going to put this here and do a
151:23 set step. We're going to use a prompt. So, we're setting this with
151:28 AI. And we're going to create a new prompt. Um, let's just set the variable
151:32 first. Let's say this is going to be put here. And this is the prompt that we're
151:42 going to be using. So, we have just asked the user to provide their name and
151:44 phone number. We need to attempt to extract the information and then confirm
151:48 it with them. Here is their reply last utterance which they've just provided
151:51 and we've captured. If there is a valid name and phone number present, then you
151:54 must do a confirmation eg okay quickly to confirm your name is this and number
151:57 is this. Is that correct? However, if one or both are missing or appear to be
152:01 invalid, you must output only retry as your response and nothing else. This
152:05 retry variable will be checked and if it matches exactly, then it will trigger
152:09 another attempt to capture. So either a write a short and sweet confirmation
152:13 message or b output retry for another attempt at capturing. So what we're
152:17 doing here is using AI to analyze the response and say look we're looking to
152:20 pluck out a name and phone number and we want to also confirm that cuz this is
152:22 likely going to be over the phone. And if the AI doesn't see a clear valid
152:27 phone number and a valid name then it's going to output only the word retry. So
152:31 give us a response. If it's good to go and we can move on to the next step. If
152:35 it says retry then we're going to try to retry. Um, now of course this retry
152:38 doesn't actually do anything unless we build the functionality in to look for
152:43 that retry keyword, which we'll do in a second untitled prompt. We can change
152:54 name. And then we can go to um go if last response which is the
153:03 variable that we're saving the output of that prompt into here. Um whether it's
153:06 going to be retry or the say just to confirm is this your phone name and
153:12 phone number. Um if last response is retry or even just to make it a bit more
153:17 flexible contains retry unless unless the person's name is like Bill Retry
153:21 Smith then uh this should be fine. And actually, just to make this look a bit
153:24 cleaner, I might change this to um if last response um does not contain retry.
153:39 here and we are going to send the last response because as we said, it's going
153:42 to either generate the confirmation message or it'll output retry. So if
153:46 it's valid, it'll and we put up last response here. They'll say their name
153:50 and phone number. We'll analyze it and then we'll go great. It doesn't contain
153:53 the word retry in it. And bang. Hey, just to confirm this is your name and
153:56 this is your phone number. And then for the choices here, we can go and use our
154:00 handy dandy. They're going to be giving us a yes or no answer to this. So, we
154:07 no. And so, if they say no to the confirmation, say no, that's not my
154:11 correct phone number or name, then we need to have some kind of retry um logic
154:16 here. I usually like to make my retries um a orange color. And then my failure
154:27 say, "Okay, let's try that again. Can you please give me a full name and phone
154:30 number, please?" And then we're going to send them all the way back to this step
154:33 here. So they're basically going to recapture their information and then put
154:36 them through this process. And this is a loop that can be done over and over and
154:39 over again. So, and then we also need to deal with this else step. So if the word
154:43 does contain retry and it has said, "Hey, look, this isn't a valid phone
154:47 number or a name," then we need to deal with that as well. So, we can come down
154:52 here and go to message. Pop it under here. And I've got a message for this.
154:55 Sorry, I didn't quite get that. Can you please give me a full name and phone
154:58 number so a member of our team can get touch. Um, right click on this and you
155:04 can go block color. Change it to an orange. And then we're going to be going
155:08 back to here as well. So, it's helpful. You can click on these arrows. So, the
155:12 lines here, and you can change them to the same color. So, we want to make them
155:16 a bit more obvious that they're coming expected. We can make this an orange one
155:24 as well. And this one, too. So, if they come in and they say,
155:30 "Hey, my name's Bill and my phone number is 02111." And it comes in here and it
155:34 goes, "Hey, that doesn't look like it's proper." It's going to send a retry as
155:38 the output. We're going to pick it up here and it's going to say, "Sorry, I
155:41 didn't quite get that." and it's going to come back up and they're going to be
155:43 expected to give it again and it will go through and then once we got a valid
155:47 name and phone number and it's not outputting retry, then it's going to go
155:50 through here. It's going to spit that out and say, "Hey, just to double check,
155:53 this is your phone number and email before we proceed." Yes and no. No is
155:56 going to be handled there and it's going to take them back to the first step
156:00 again. So, that's some nice um error handling and sort of looping that you're
156:02 going to need to be building into a lot of your conversational AI agents,
156:05 especially on voice flow, right? And so, the last steps here are some quick
156:08 variable extractions. So we can go to div logic here and go to set and we're
156:13 going to extract the name and the phone number. So we'll go prompt. Holy moly,
156:19 it is it's bloody hot here. The variable that we want to save this, we were going
156:23 to be extracting the name. So we'll add name. Um, and this is going to be called
156:37 name. Here's a quick and easy prompt. You can pause it to take a look at that.
156:39 It's just going to extract their name. We need to add another variable.
156:53 prompt. Just paste this one in here. Pretty basic. Again, pause it if you
156:56 want to take a look. And I haven't named that prompt. It's going to annoy
157:05 me. Extract phone number. And now we're going to have their name and their phone number
157:11 extracted out of this response. And oh, and actually we need to add in
157:18 the conversation history there so that go great. Let me get that added into our
157:35 system. This buys us a bit of time as we use our um make web hook which we're
157:39 going to set up now. So the next step is to get a Google sheet set up and to use
157:44 make.com to uh take this data and shoot it into a Google sheet. So to do that um
157:48 I will leave a link on I mean you can just search it up. It's make it's
157:52 make.com. All right. I'll save you the hassle. So you sign into make.com create
157:57 an account whatever you want to do or Go to scenarios and then we're going to
158:08 create a new scenario. I'm going to build from scratch here. I'm going to get all that
158:13 rubbish out of the way. I don't know why it's acting like I'm some rookie here.
158:20 Um, and then we need to go to web hooks, custom web hook. We're going to
158:26 um we're going to add in a new web hook. This is going to be conors cleaning lead
158:32 capture. going to save that. It's going to create this web hook here. I'm going
158:37 to copy this edges to clipboard. We're going to come back to our build and then
158:41 we're going to go to the API step. So, what this is doing if
158:45 you're a bit new to to web hooks and and API calls and stuff. What we can do here
158:50 on make is set this up to basically listen. It's a URL. You know how we had
158:54 the endpoint? The end point, this thing here that it's given us that I've just
158:58 copied to our clipboard. That's like the address, remember? So if you put it
159:01 write it on the on the letter, that's where it's going to go. This allows us
159:05 to basically send data um via API call um from voice flow to make and it's
159:09 going to catch it here. And this little lightning bolt means that anything we
159:13 build after is going to be triggered whenever it receives one of those uh
159:17 whenever a new bit of mail arrives. It's going to then trigger this multi-step
159:21 process. So we're going to put this into voice flow. We're going to trigger it
159:24 and make sure it knows what data to expect. And then we're going to be able
159:27 to use that data and put it into Google Sheets. So stick with me here, but this
159:31 is another very very essential skill in AI automation is how to set up a webbook
159:35 um and use it within different apps. So we have our webbook here. We've copied the address to
159:41 clipboard. We're going back setting up a get request here. So it's not a post
159:45 request this time. We're just uh getting and we're technically not getting data.
159:48 A get request is a much more kind of quick and dirty request. Um and as
159:53 you'll see, we kind of just tack on a bunch of information after this. Um we
159:56 can do it through what's called parameters. Here we want to be sending a
160:02 property. Oh, let's just say property type. Actually, let's do this properly.
160:09 Let's go time um timestamp. So, in this Google sheet, you're going to want to know when the
160:16 different leads came in. So, we can go uh time stamp and get the time stamp
160:20 from voice flow. That's a default variable that they are automatically
160:23 filling out for you. So, one of the things, one of the rows in the
160:27 spreadsheet is going to be uh the time stamp. We're going to add another and
160:32 we're going to go um name. We're going to put in the name here. So, now all the
160:36 cogs in your head should start turning as we put this together. And we're going
160:44 to go uh phone number and we go bracket phone number number. And if we add another, we can go
160:52 property type. We can go size and go property size. We can add
161:12 another one. We can go quoted prices so that the sales team knows what we actually told them. in
161:18 case you're playing around with pricing. Um, raw quote data. And I'll probably
161:24 throw in one more here, which is their uh, user first reply. So, that might
161:34 give context to the sales team like what did they actually contact us for in the
161:37 first place? Maybe helpful, maybe not. But we can just send this all over to
161:47 And so now we can see this is uh as this thing's spinning around, it's basically
161:50 waiting for us to send some data to it. It's basically sitting there at the
161:54 mailbox like waiting for it to come through. Um and we can go send here and
162:00 I'm going to put in uh gosh dug myself a hole here. Um let's go name
162:09 Liam. Um house probably size. Um um lots of money. Um how much for
162:13 cleaning yarning? um send. And if we go back, bam, successfully determined. And what we've
162:20 done and determined means is that make has received the the the request that we
162:24 sent. And it now it knows that we're going to be sending it a time stamp and
162:29 a property type and this this really really key skill to understand because
162:33 now when I go oh save now when I go to here and I go Google Sheets and I go add
162:38 a row, um you will need to set up your uh Google Sheets connection here. So, you
162:44 just sign in with Google, add your connection in. Um, I am going to have to
163:00 quickly. I'm going go timestamp. So I'll zoom this up. Time stamp name
163:06 phone size. 1 2 3 4 5 6 7. Yeah, we've got them all. Right. So then I can
163:26 go. So call this my Connor cleaning support agent leads. So now I want to go back to make and I
163:37 want to click here to um Connor. There we go. Connor's clearing. Why do I keep spelling
163:49 cleaning? Now we have the spreadsheet set up. The sheet name is just going to
163:54 be sheet one. Sheet one. Does the table contain headers? Yes, it does. And now
163:58 we get to put in all of our values. So, bam. Time stamp. Pop that in there.
164:04 Name, pop that in there. Phone, pop that in there. Property type. So, you see
164:10 what such a key skill that uh I really, really want you guys to learn. Um,
164:14 because this is a lot of like if you just have a couple of these key things
164:17 using these APIs and say voice flow or in NAD, building tools and relevance and
164:22 then using them via API, knowing how to set up web hooks and then send data in
164:25 between them and to set this data structure. You send that first initial
164:29 batch over to make in a test test request. It's going to go okay this is
164:32 what they're going to send me in future and then that's locked in and you can
164:35 send thousands and thousands of requests through it and it will operate as you
164:38 expect and first question. So uh it's got all those other rows we don't need to worry about. Save.
164:44 All right. And so that should be working. We can switch this to
164:49 immediately. Oh, I need to save it. Save. Um I'm going to immediately as data arrives. So, this
164:55 is going to be waiting all the damn time. Actually, I'm I've got a whole
164:59 bunch of other stuff running in make for my my various businesses. So, I'm just
165:01 going to have this set up. If you wanted this to run around the clock, you turn
165:04 that on. I'm just going to turn it on to run once. And this is just going to be
165:07 sitting there waiting for data. So, what we can do is just finish this off
165:13 quickly. So, I'll just paste this in. Got this all done. Thank you. That's
165:19 the end of it. Um, what we want to do is have a failure path. So, in case this
165:32 say, "Darn it, something went red. Um, and then this one is like that.
165:42 All righty. So, if we look at make, it's still expecting us to send some data to
165:46 it. Let's hope that we've done all of this right and it's going to work first
165:50 time, but you never really know, dear. So, let's give this a run and we'll try
165:55 to go through that second path and we'll spreadsheet. I need weekend cleaning,
166:13 So asking a question is going straight to price. You're looking for weekend
166:16 cleaning services and want to know the pricing. Is that correct? Yes. So our
166:19 router is going to pick that up and say he's looking for pricing. Let's send him
166:22 to the quote tool. Okay. Sure. To give you an instant quote, I just need the
166:24 property type and size and square feet. It's a property. Let's go aart
166:30 mint. It's detected that correctly. And typical apartment size
166:47 feet. Okay, so 500 isn't actually ridiculous. Um, great. And then we get
166:50 our answer based on your 500 foot apartment. Bam, bam, bam. There's all
166:53 the prices. Please provide your name and phone number and I'll get one of the
166:55 teams to call you with a time that works. Let's uh give this a a test. See
167:00 if it's actually picking up on the fail. Sure. My name is and
167:14 is boom. Sorry, I didn't quite get that. So, it detected that it wasn't right.
167:18 So, we got the retry output. We got the retry output from this, which is what we
167:21 wanted. Sorry, I didn't quite get that. Can you give me again? Um Liam
167:36 021. That's what numbers look like here in New Zealand. Okay. Just to confirm quickly,
167:41 your name is Liam Mley and your phone number is that correct? Yes, sir. Name
167:50 number. Oh, what has it done there? Oh, dame. I don't know why it's I just want the
167:57 phone number. So, we have got a little bit of an error there. I just go back
168:00 and tweak the prompt. Make sure it's like only get the phone number. We don't
168:03 want anything apart from numbers here. Great. Let me get the added to the
168:06 system. And then if we go back to make, we see these are all green now. And we
168:09 see these dots. So, this is the information that came through. Namely,
168:12 Mly phone number. And so, here's the little mistake where we had a new line.
168:16 Apartment. Bam, bam, bam. All of that information. And then it's added it into
168:21 Google Sheets. Here we see updates, updated number of rows, all the values
168:32 to. Okay. Holy moly. Right. We're ready to put this thing on our website and to
168:36 also put it on a phone number. So, let's just finish the job. Guys, I'm uh
168:40 getting real hungry, but we can uh we can push through. So, I will just turn
168:44 this on actually so that if we are testing it on the web and over the
168:49 phone, um it's ready to receive. Um, if we do want to be pedantic, I would go
168:54 back and I would change I have to do it. It's going to piss me
168:59 off. I'll put their name. Oh, that's why. I'll put their phone number only.
169:14 only. And we do have these fail points here. Um, I'm not going to bother
169:17 filling them out. I think you guys can figure out based off how I've handled
169:21 this, how you can handle these as well. So, what you'll find is when you're
169:24 building these, these kind of fail like error handling um is kind of a an
169:28 endless thread that you keep pulling. It's like, oh, well, now I've got to
169:31 handle this, this, and this this. So, um I'm not going to this is a prototype.
169:34 I'm not going to be doing all of the the error handling for you here. In the
169:37 template, there is actually a little bit more of it. Um some better examples. So,
169:41 maybe if you import that, you can just steal the work that I've done there. But
169:45 what we need to do now is we can publish this thing. We'll call it
170:03 drop. All right. So now it's published. We can add the agent to a website. Let's
170:06 click on that. And that takes us to this integrations tab. Um let's put this down. I don't need to
170:13 see that. Um, they've got a new version of it. That's good to know. I said
170:16 installation is pretty straightforward. So, we can just click copy here. And
170:24 up I'm going to open up brackets here just to give you a demo of a website. I
170:27 use this in all my tutorials. It's really easy to spin up. Um, I will leave
170:31 a link to this template if you want it. Um, and also some instructions on how
170:34 you can open up a website. I know this looks like code and it's all scary, but
170:38 um, this is just allowing me to spin up a website very quickly. So, I'll leave
170:42 the template file. All you need to do is once you've downloaded the template
170:46 file, you need to download brackets, which is the software. You can go file,
170:50 then open folder, and then you want to click on the folder when you've unzipped
170:53 it, and it's going to open up the whole folder. And then you'll get all of this
170:57 uh opened up like this. And you see all of these files ra. All you need to do is
171:00 click on the index.html. And then you'll see something similar to this. Well, I'm
171:05 going to scroll down to the bottom of the index.html. I'm going to delete this
171:10 old voice agent I was testing on here. Drop this in here. Paste that. And then
171:14 save it. Command S. Click this little button up here. And it will show us a
171:20 local version of the website running on our computer here through brackets. So
171:24 here's my man with a magnificent beard. And we have the tester agent bubble down
171:40 I want to know where you Yep. Woo. Okay. Uh I didn't even program
171:54 that in there. Maybe you just thought it was inappropriate. Um but we've got it
171:58 working on a website. Now, if we pop back over to um uh Voice Flow here and
172:04 you go to the integrations, the widget, you see we've got this test your agent
172:07 thing. So, down here, we can play around with the look and feel of it. I'm not
172:10 really going to get into this here. There's quite a lot to play around with,
172:14 but basically all of what you see on here can be changed around. Different
172:18 logos, different text here, different icon, etc., different colors, and you
172:22 can just make it look and feel however you want it to. So, I'm sure you guys
172:25 are big enough and ugly enough to figure that out yourself. we'd probably want to
172:29 switch over to uh this here. One thing you would want to do is turn off powered
172:33 by voice flow so it's not uh sending traffic to them when it's on your own
172:36 website. And that's about it. For the sake of time, I'm not going to go
172:40 through the entire flow again here. Just know that the functionality that we
172:42 built that I just showed you in the builder is going to work cuz we just
172:45 deployed it. Like this is exactly what we're interacting with. So, it's all
172:49 working here. The only step to do now is to put it on this phone number so we can
172:52 have a chat to it over the phone, which we're going to do now. To do that, we
172:55 need to go to the telefan bit here. It is in beta right now, but for most of
172:58 you watching it is not going to be by the time you you are watching this. So,
173:01 we need to set up a phone number from Twilio, import it, and then connect our
173:04 agent and its functionality to that. So, we can go import number. You'll see that
173:08 we have this information here. So, we can use Twilio or Vonnage. Twilio is
173:12 usually the go-to here. So, if we click on learn more, then they're going to
173:15 help us. Basically, the best way to make sure you're getting the most up-to-date
173:18 information is go to the docs of the platform. Finding and reading and
173:20 extracting information from documentation on these kinds of platforms is another key skill that you
173:25 need to pick up to succeed in the space. So if we go to the docs here, we go to
173:34 um voice phone number setting up Twilio integration and they have a video here
173:37 adding a phone number to your agent. So if you ever get stuck, you know, you've
173:40 got documentation here and for all of the other platforms, but they'll keep
173:43 updating these videos if things change, which they likely will as this voice AI
173:48 space really takes off. So, if we go to Twilio, you will need to sign up and
174:06 Twilio. All right, so we are logged into Twilio. You'll need to create an account
174:10 for most of you, but Twilio is a uh phone number provider that you can
174:13 connect to and interact with over the internet. super helpful when you can buy
174:16 lots of numbers from different locations and stuff. When it comes to phone
174:19 numbers, it can there's a lot of rules and regulations around different like it
174:23 varies a lot from country to country. So depending like if you're in Germany, I
174:26 believe in order to get a German phone number, you need to have a company
174:29 registered and get the number through your company registration and provide
174:33 those details. So can be difficult. I'm just going to show you how to use a uh a
174:36 US-based number here. So we can go over to the phone numbers on the left here.
174:40 There may be some setup that Twilio walk you through. It can be kind of annoying
174:43 sometimes. They say you need to do all of these declarations and forms and
174:46 stuff, but for the most part, it should be fairly straightforward if you follow
174:49 their setup instructions when you create your account to then come over and go to
174:54 your phone numbers and manage and go to buy a number. Now, unless you have other
174:59 purposes you want to use this for, you can just snag any random one if you're
175:02 following this tutorial. Um, if you're obviously doing this for a client, you
175:04 could get one that's matched to their location or their their state or even
175:08 their city. And when you click buy, you can see there's all these kind of
175:11 registrations and RAR you need to do. But thankfully, voice is uh is ones that
175:15 don't need all of that. And you've got global routing, etc. So, you can come
175:18 down here and buy this. It's going to be a dollar a month. I know cost of
175:21 starting up a business is ridiculous these days. How dare they? But just to walk
175:26 things through and do it with you, I'm going to buy this number even though I
175:29 really don't need it. There we go. I've got my number. I'm just going to copy this so I don't
175:39 forget it. 517. And then I go to active numbers. I got my guy. Where is he? 517
175:44 down here. Copy this bad boy. Go back to Twilio. Put our number in here. Our
175:51 Twilio account SID. You need to pop back to Twilio. You go to your account
175:56 dashboard. This was a bit confusing when I first did it. So, I'm glad I'm glad I
175:59 can help you guys through it because it was a bit of a pain in the bum for me.
176:02 You think all this information is going to be like tucked away in the side
176:05 panels, but no, it's just on the dashboard. Who the hell puts your or
176:09 tokens and stuff on a dashboard? I mean, it's when you know it's there, it's
176:11 helpful, but it was a bit confusing at the time for me. Um, I'm going to paste
176:22 here. We're going to copy this or token and paste this in here. And we're going
176:29 to say this is my you YouTube tutorial. And I'm going to import that bad boy.
176:33 Select environment. I'm just going to put it on development for now. uh we can
176:36 assign that. And with that, you can see if I click unassign here, that means I'm
176:41 going to take it off. So, it's it's assigned and it should be ready to have
176:44 a chat with. So, if I click to copy this number here, I go back to my little
176:52 uh uh website code here. If I go find um plus one. Oh, there we go. So, we have
176:58 the call. Um this is in the header. Uh this is just me p wanting to put it on
177:01 the site for you guys. So, you guys don't necessarily need to copy this. I'm
177:05 going to save that. If I go back to my website here, now I have the actual
177:10 phone number. And if I pull up Skype on my phone, you guys, if you're in the US,
177:13 you can just or wherever your number is based, you can just call it stage plus
177:20 one. All right, let's give our mate a Welcome to Conor's Cleaning. How can I
177:28 help? Hey, uh, I was just wondering, uh, business. Is that correct? Yeah, that's
177:52 We're located at 247 Beacon Street in Boston, Massachusetts 0216. We serve the
177:57 greater Boston area, including neighborhoods like downtown Boston,
178:01 Cambridge, Somerville, Brooklyn, and several others. Is there anything else I
178:05 can help you with today? Sweet. Okay, so that's done the uh knowledge the knowledgebased route.
178:12 Now, let's give it another spin. Call Hey, welcome to Connor's Cleaning. How
178:20 can I help? Hi there. I'm just getting interested in getting uh some cleaning
178:30 It sounds like you're interested in booking a cleaning service for a
178:43 Okay. Sure. To give you an instant quote. I just need the property type and
178:47 size and square feet. Is the property a Sorry, I didn't get that. Is it a house
179:08 And how many square feet is it? It's One sec while I generate your quote.
179:20 Based on your requirements, we can offer four different packages. Regular
179:24 maintenance cleaning at $90, which covers standard weekly or bi-weekly
179:30 cleaning. Deep cleaning at $150 for a more thorough top to bottom refresh,
179:36 move in, move out. Cleaning at $170 for preparing or cleaning after a home
179:40 transition, and post renovation cleaning at $200 for handling construction dust
179:45 and debris. Please provide your name and phone number and I'll get one of the
179:48 team to call you to find a time that works. Sure. My name is Liam Otsley and
180:05 Okay, just to quickly confirm, your name is Liam Oxley and your phone number is
180:11 02119792. Is that correct? Yep, that's Great. Let me get that added into our
180:24 system. All done. Thank you. Boom. All righty. That is done. That is dunzos.
180:32 All right. So, you guys learned a lot in that one. voice integration, website
180:37 integration, um connecting web hooks to make, setting up make automations,
180:41 sending data over, connecting relevance AI tools into into voice flow. Um basic
180:47 integration with a CRM, in this case, it's a sheet, but there's so much in
180:51 there, guys. I hope you really really uh learned a lot from that. This has been a
180:54 big one. And we've still got uh one more to go. So, I hope you're sticking with
181:00 us. Um but going back to our Figma here, um we have ticked off all of this. So,
181:03 we have it as a web chat widget and we have it as a a phone number. Now, as far
181:07 as I know, you can have both options for the same agent on voice. You can have it
181:10 on the website and over voice. You don't need to duplicate it and sort of define
181:14 what modality it's going to be. So, we've ticked off all the boxes for this.
181:17 All of the resources will be in here. All of the prompts, a template for that
181:20 whole final build as well. If you just want to snag all my hard work and go and
181:23 sell it to someone, again, I don't I really don't care. Um, that's what these
181:26 videos are for. And we're getting into All righty. So, last but not least is an
3:01:29 Build 4
181:38 agent built on my own software. So, I didn't want to make this. It's not about
181:41 me selling you or getting you to use my software. So, I thought I'd put it at
181:43 the end just so you know that I wasn't really This is about you guys learning
181:47 and my software happens to help you put an agent onto WhatsApp very very easily.
181:50 So, that's why it's included in here. But again, this is nonsponsored,
181:54 nonpromoted, non whatever. I'm just really trying to share with you what I
181:57 think is a really valuable skill set to have. All right. Now getting into AI
182:00 agent build number four. This is going to be tada a WhatsApp based ARI customer
182:06 support and lead generation agent built on agentive my software. So this is a
182:11 noode uh AI agent builder that is built on top of OpenAI's assistance API. So
182:15 you're technically using your OpenAI account getting very very cheap rates on
182:19 the uh token usage that you're running through this agent. But Agent just
182:22 allows you to build on top of it very easily. but more importantly to deploy
182:26 these agents not just onto web chat widgets like we've done with voice flow
182:30 but easily onto things like WhatsApp and Instagram etc. So that's really the key
182:33 thing that Agenda focuses on doing right now is making it easy for you to get
182:36 your agents onto these platforms. So as you can see it's a fairly similar build
182:39 to what we just did on voice flow in terms of functionality. We're going to
182:43 be having a uh a knowledge base that we can ask questions over. It's going to be
182:46 able to generate another instant quote. So we'll just quickly connect that same
182:50 relevance tool here. And finally, we're going to do a lead capture, but this
182:53 time it's going to be done through Air Table. So, I want to mix it up and show
182:55 you how you can connect your agents to Air Table, which is a very, very common
182:58 integration that you're going to need to know. And the difference between this
183:01 agent, you're going to see that it's much much faster to build. This is not
183:05 meant to be a side-by-side comparison of what's better, how much faster. It's
183:08 just that when you build on a more conversationalbased uh AI agent platform
183:12 like agentive which is built on top of the assistance API, it's a very
183:15 different way of building agents because it's all just based on a prompt and
183:18 providing the right tools and all the magic kind of happens itself through the
183:22 prompt. Whereas voice flow gives you a lot more control. So it's really
183:25 difference between structured AI agent building versus more conversational and
183:30 open-ended chats through more chat GBT like experience that can just go on and
183:33 on and on which is what these agents can do. So the purpose is of course fairly
183:36 similar but the value of this is slightly different in that we are using
183:40 WhatsApp. So uh many people browsing for services online are hesitant to use
183:44 website contact forms or other or chat bots that they think are not going to
183:47 give them access to a real human due to the potential delays that come from it.
183:50 Right? You land in a website and you're you're shopping around for a different
183:53 service or product and then there's this this contact form or there's a a web
183:56 chat widget and you're going to go well I don't really think I'm going to get
183:59 the help that I really need here at the at the speed that I want. So you might
184:02 look for a WhatsApp widget and you know that if you click that WhatsApp widget
184:05 you're going to get to speak directly to someone and this is kind of playing on
184:08 that fact that if you have the WhatsApp option on your uh website people are
184:11 much more likely to just click that and go through and try to have a
184:14 conversation directly to get what they want. So by having a WhatsApp option on
184:18 a website or other triggers eg you can have a QR code that you could stick on
184:21 say a real estate sign and you build an agent connect it to your WhatsApp number
184:25 like we're going to do here and then you create a QR code that people can scan
184:28 and immediately open WhatsApp and start chatting with it. There's lots of
184:31 different ways that you can have an access point into a WhatsApp agent like
184:34 this. But it basically opens up more conversations through a more smartphone
184:38 native platform. So they can hop on their phone and sort of have a chat away
184:40 to it rather than being on a website on the computer or a little tiny website
184:45 chatbot on their phone. Um, in order to essentially engage more prospects or
184:48 more people interested in the business in conversation, quickly provide value
184:52 through either the knowledge base and these tools here and real-time quotes.
184:56 And ultimately, because you're providing that instant value and instant feedback
184:59 from them, collect their lead information, or better yet, even set
185:02 appointments through WhatsApp, which you can build on agenda. But that use case
185:05 is a little bit more advanced and not something I can show within this video,
185:08 but it definitely is possible. But it's really only a few steps away from the
185:11 skills that you've learned in this video so far. So keep an eye on that
185:14 appointment setting use case because if you can do that with AI agent, it's a
185:17 very, very valuable one. And I've done other videos on the channel here showing
185:21 you how to do that. So, the usage of this is that they're going to find the
185:24 uh company's WhatsApp number on their website or elsewhere, maybe a QR code
185:27 like I said, and they're going to start a conversation on WhatsApp, and then the
185:30 agent is immediately going to jump in and start responding and be able to
185:33 answer from the knowledge base, generate quotes, and then capture their lead
185:36 information. So, without further ado, let's jump into building this agent. So,
185:39 we can click up here to go to my website, Agentive. You can click on
185:42 register now. You can just register with your Google account. I'm going to log in
185:50 It is free to make an account and we have a free plan so you can just
185:52 experiment around as much as you need and then you're only going to be charged
185:55 based on the amount of usage you use. So it's very very cheap and affordable and
185:58 I wanted to make this platform for you guys to all get on and experiment with
186:02 building AI agents without coding. That was really the the core of why we
186:05 started this whole thing. So we've got the dashboard here which will load in my
186:07 data in a second. So you can see here what the dashboard will look like when you've got
186:12 your own agents running. We are running the Agentive customer support chatbot
186:16 through this uh through this account that I'm I'm using right now. So you can
186:20 see usage costs very cheap sessions etc. So it's really cool when you go into
186:23 analytics and you can use agentive to see how people are using your agents but
186:26 that's obviously something for a little bit later once you put these into
186:29 production. Uh now what we're going to do is of course we can go to agents or
186:33 we can just create an agent from here and I call this um Connor's cleaning
186:39 WhatsApp agent. Oh we got a little description as well. um answers
186:50 time so answers questions from the knowledge base provides real-time
186:54 cleaning quotes and can capture leads to air table. So you're going to see the
186:57 setup is a lot faster than some of these other platforms. Again, I'm not not
187:01 trying to gas myself up here. It's just a different way of approaching u
187:04 building agents. So it's a lot more fast and and rapid prototyping and easy to
187:08 get things up and running. Of course, if you need much more advanced
187:10 functionality, you do need to go the extra mile and go on to platforms like
187:13 Voice Flow. But in this case, we have a prompt, very easy. We have a knowledge
187:16 and we have tools. So, remember when we went back to being a a chef and the
187:20 three ingredients concept, these are your three ingredients, right? The
187:23 prompt that you get to provide as the instructions, the knowledge that you
187:26 provide as the external knowledge base and the tools that we can connect to it
187:29 as well. And we can select the model here. So, I think I want a nice and
187:33 snappy response time because this is going to be on WhatsApp and customerf
187:38 facing. So I'll go to GPT4 mini. So it's nice and quick. Now we can just put a
187:46 help. Just put that in there for now as the prompt. The knowledge base we can
187:50 turn this on. We can create a new this. We're going to click here and
187:59 we're going to upload that same file that we used on voice flow. The same
188:01 document that will be available in the resources for this uh for this
188:05 particular guide. Give that a second to process. Once this goes green, we're
188:07 good to upload it. And you can add multiple files in here. We allowed five
188:11 files at a time, but you can add dozens and dozens of files. So, you have a
188:17 with. And just like that, we have connected our knowledge base. And the
188:21 cool thing about Agent is because we're built on the assistance API, this is
188:24 actually an independent knowledge base. So, you can create a knowledge base and
188:27 connect it to multiple different agents. The the knowledge is not restricted to
188:30 the agent that you build it within. So, I can go and create a new agent and
188:32 connect this exact same knowledge base. And I have all of these other ones here.
188:35 And then when we go into the tools section, we are going to have two tools
188:38 for this. Well, the knowledge base, if we go back to our Figma here,
188:41 technically the knowledge base is a form of tool that the agent is using. But on
188:45 platforms like the assistance API and and many platforms, you'll see knowledge
188:49 treated as its own thing, but essentially it is just another tool that
188:52 the agent is using at the right time when it needs to pull in knowledge to
188:56 answer questions. So, OpenAI separates it out into its own thing here. And so
189:00 do we because we built on top of it. So we do have three different tools but
189:03 knowledge is its own tool that gets set up through this knowledgebased um
189:07 connection that we just made before. Then we have the tools and here we have
189:11 our instant quote from relevance and we have our capture lead information. So we
189:14 know the process of going on to relevance. So we can just go create a
189:17 new tool here and this is going to show you a schema. Remember back to when we
189:21 talked about schemas it explains to the agent how to use the tool. So to add a
189:25 tool to this agent we need to add a schema to it. And thankfully our buddies
189:29 at relevance AI provide a very very easy way to create schemas to import into
189:33 agents like on aentive. So here I can grab that same cost estimate tool for
189:36 the instant quotation for cleaning services that we've used previously.
189:39 Again this will be linked. You can just clone this if you haven't got it
189:42 already. I will provide a link for you to clone this into your relevant account
189:45 that will be in the resources for this video. And it's just like the previous
189:48 tutorial that we did where it's got property type square footage and an LLM
189:52 step here to calculate it. It's going to spit that back and we're going to turn
189:56 that into a nice response uh with our agent on Agent. We can make sure that
189:58 we've saved this. So, the cool thing here is that in order to get this
190:02 connected to uh Agent and our agent over there, we can just go to custom actions
190:06 here on the tools page. As you can see here, it's mainly intended for use with
190:10 OpenAI's custom GPTs, which you can get access through chat GPT. And I highly
190:14 recommend you do check out the OpenAI GPTs because it's a super simple way to
190:17 spin up your own agents um on the chat GPT site. And so, we can select our tool
190:21 here and we can get a schema for it. But what I've just realized is that I
190:24 actually do have an air table lead capture tool here that I've already
190:26 created on relevance. And it's actually going to be easier for us to set it up
190:30 here on relevance than to have to do it all separately. So let's just quickly
190:35 set that up. Now if we go air table, let's just get a simple one that
190:38 captures the name and phone. I'll provide the template for this tool so
190:41 you can clone it in. But basically it takes an input of the name of the lead
190:44 and the email address of the lead and the phone as well. So, it's capturing
190:48 all three of these as lead information and then it's sending it over to Air
190:51 Table which we're going to set up just now and it's using a post request to
190:55 push that data that we collected here and we will collect through WhatsApp
190:58 eventually and it's pushing it into the Air Table database. So, let's get that
191:07 Table and I'm just going to use a dummy CRM that I use for all of these
191:10 tutorials and you guys are going to be able to clone this if you want. In the
191:12 resources for this video, there will be a link like this. So, if I share this um
191:20 publicly, you guys will get something that looks a bit like this and this
191:24 button up here says copy base. That will copy it into your account. So, all you
191:27 need to do to copy this air table base is to create an air table account and
191:30 then click on this copy base and it will copy it over. So, you can get this with
191:34 the column source preset up. It is fairly easy to set up these fields
191:36 yourself, but I want to make it easier for you guys. So, you can just copy
191:38 this. It'll be included in the resources. But here we have the fields
191:42 that we're looking for. So now all we need to do to send data into this
191:45 database through our WhatsApp based agentive agent. So when someone provides
191:48 their details that it gets shot into here is we need to go and see our
191:52 details for the air tableable web API. So air tableable has their own API which
191:56 allows us to interact with our databases like this programmatically. So all we
191:59 need to do is go up to the right hand corner here go to builder hub and if we
192:04 go to the developer docs here and scroll down to the web API this is a reference for the air
192:10 table web API. So this documentation is essentially going to tell us how to
192:13 interact with our air table programmatically through our agents and
192:16 through voice flow and through relevance and and through agentive as well. So any
192:19 way you want to interact with it, you can now take this knowledge that you've
192:22 gained in this video look through this and model what we're going to do here in
192:25 relevance. You can take that same idea and put it into say voice flow and you
192:28 can build an air table integration within voice flow yourself where you can
192:32 send and pull data. So these skills all stack on top of each other and it really
192:35 centers around understanding how APIs work and that comes down to reading
192:39 documentation as well. So in this case, if you go back to our relevance tool
192:44 here, we need to get our URL, which is our endpoint, which we've talked about
192:46 before. This is the address that we are sending the request to and agentive
192:50 where we're building the agent. It's going to be using relevance to call air
192:53 table. It's a bit of a a roundabout way of doing things. But to get all of this
192:57 information, the easiest way is to go back to this documentation. And the
193:00 easiest way for us to find the information that allows us to interact
193:03 with our own Air Table base that we're setting up, is to come down here and
193:07 find the base that you've just cloned into your account, which will likely be
193:10 Smith Solar CRM. Don't worry about the name. And Air Table does a really,
193:13 really good job of making this super easy. And that we can just come here to
193:17 the leads table. So, if we go back to um Air Table here, you see we are on the
193:21 leads table. We have these different tabs. You can just ignore these are just
193:24 different projects that I've done on YouTube. Um they're all kind of there in
193:27 case people are also cloning this into their account. But we're looking at the
193:30 leads tab here. So, we go to the leads here and we want to create records. And
193:34 then it gives us all of the information we need here in order to create records.
193:38 So, um you can see that it's a post HTTPS and all of this information. So,
193:45 this is the endpoint. We want to copy all of this all the way down to leads.
193:50 Copy this. Go back to relevance and paste this in. Oh, maybe
193:55 that was already there. And paste that in there. And then we need to add in two
193:59 headers. So we have authorization and content type. Now remembering what we
194:02 learned before, you can see we have H and this H tag means that there's a
194:06 header. And so the header is authorization. And then the value is
194:10 going to be bearer and then our token. This is something that tripped me up
194:13 when I was first learning this uh using APIs. Is that you need to add this
194:18 bearer word and then a space and then your API key. It's a weird way of doing
194:21 things. I don't really know why uh why it's like that but sometimes when you're
194:25 doing these authorizations you need to add in bearer space and then your API
194:29 key. So it's it's there for a reason um is what I'm saying. And then we have the
194:33 content type being application JSON. So we're already familiar with that. So
194:36 going back to relevance we have the header of authorization and content type
194:40 here. Application JSON. And now we need to add in our Air Table API key so that
194:44 we are authenticated and we have permission to send an API request. So
194:47 they're not going to let anyone use this details and and start sending data to
194:51 our our database, right? They need to be authenticated and that's what API keys
194:55 do. So to get our Air Table API key, of course, we go to Air Table. We can come
194:59 up to the top right here, go back to our builder hub, go to personal access
195:06 tokens, and we can create a new token. Call this YouTube. We can go um add the base. This
195:15 will be um Smith's Solar CRM. We can add a scope read write and sometimes I find it handy
195:22 to have the schemas read in there as well. So basically what we're doing here
195:25 is saying that I give this API key that we're creating permission to interact
195:29 with this uh this air table and I give it permissions to do these things like
195:33 read what's in the database write to the databases and create new things and also
195:36 to see the overall structure of the base and the field types. So we can add that
195:41 and create the token. We get this token, head back to relevance. And again, this
195:45 template so that you can clone it into your account is going to be on the uh on
195:48 the resources. So, if you're following along, you should just clone it into
195:51 your account and then come down here and make the changes as I do them. So, we
195:54 can add a make sure we have a space after bearer and then paste our key
195:57 because if we go back to the web API docs, we can see we have authorization
196:02 bearer space your API key content type and then application/json. Then we have
196:05 the data as the payload. Remember, like this is what's inside the envelope. Then
196:10 we have records and we have the fields name, phone, and status. And then it's
196:13 provided us an example of how we would send data into that which we don't need
196:16 to worry too much about because I've already got this fitted in here. It can
196:19 be quite fiddly. In fact, for this I'm actually going to add another field in
196:23 here. This is one thing about relevance I'm not a huge fan of. This can feel
196:27 super fiddly sometimes. So if we go email, this will already be in the
196:34 template that I give you, by the way. Okay, so we have the URL has been
196:47 updated. The method is post. That's correct. We have the authentication. We
196:51 have the content type. We have the body all set up. We've added in our fields of
196:56 name, phone, and email, which we have name, email, and phone. So, we can give
197:01 it a spin here. If we say and we give it a spin. Run the us. Yep. And there we go. If we go back
197:22 to Air Table, open the base up. There we have it. Liam phone email.
197:29 So, we can take this tool and we can take the instant quote generator. we can
197:33 put those into uh Agent and before you know it, we're going to have our agent
197:37 ready to go. So, let's head back to relevance here. We'll save this tool and
197:41 we'll change this to name, email, and phone. Um, and just quickly
197:46 before we do that integration, this is when the description comes into play
197:49 here. Remember those natural language descriptions of what the tool does, what
197:53 each of the parameters and inputs are. It's really important to get these right
197:56 in relevance, and I see a lot of people skipping over this step, but this is
197:59 what's going to be put into that schema, right? So when relevance generates a
198:03 schema for us, that onepage manual on how to use this tool and use the API in
198:07 order to interact with this functionality when we give it into
198:11 agentive, it's going to be reading over everything in there. And it's going to
198:13 be those little descriptions around what the tool does and what it's supposed to
198:17 take in. And and these parts here in relevance is where we get to set that
198:20 up. A proper description is needed before we do this integration. So this
198:23 tool captures lead information, stores in Air Table CRM, requires lead's name,
198:27 phone, and email. Name, phone, and email. The name is name of the lead.
198:31 Yep. Email, email address of the lead and phone is the phone number of the
198:35 lead. So that's all good there and ready to integrate. Might even do a quick check
198:47 well. Yep. Type of property, square footage of the property, and we are
198:51 ready to go. So now we can click on the custom action step here. Scroll down and
198:59 click on both of these. Bam. Bam. Scroll down. We're going to change this
199:03 to custom orth. We're going to generate an API key. There we go. And we're going to
199:10 generate our open API, not open AI, open API. It's essentially a type of API and
199:14 a way of describing how the API works. And it gives us all of this information
199:17 here. I will actually expand it out so you guys can see at least some of it.
199:21 It's probably easier over on agentive actually. And then we head back to
199:25 agentive. What we can do is paste in the schema. And now if we scroll through
199:31 this quickly, I just want you to see what a schema looks like under the hood
199:34 because we have some important parts uh that's going to really connect the dots
199:37 for you after everything that we've learned in this video. So I'll zoom in a
199:41 bit here. Um we have the title of the tool. So we have a few key things in
199:44 here that we can break down. Basically the paths. We have two paths in here.
199:48 This is one of them and this is the other. These represent the two tools
199:52 that we are integrating. You can see one here is the operation ID is basically
199:56 the name of the tool and that is taken from relevance directly the air table
199:59 lead capture and the summary here is the actual name or the title of the tool
200:02 that we had in relevance. This is just a a version where they put in um
200:06 underscores to connect the uh the gaps and the description here you can see
200:09 it's the same as a description that we set up over I don't want to go back on
200:13 there but that was a description that we put under the name to describe what the
200:16 tool does and then as for the inputs relevance has made it a little bit more
200:19 complicated by putting a schema in here. Um, so we'll cover that in a second, but
200:22 basically here's the second tool. Sparkly cost estimate. This tool does
200:26 this. This about estimating the cost of an apartment. Then down here we have the
200:31 schemas for the inputs. So we have things like the name. This is for the
200:35 lead capture tool. The name um this is one of the fields. It's going to be in
200:38 type string and it's required. We have the email which is type string which is
200:42 required. And we have the email which is a description here. And of course you can see all the
200:47 descriptions that we put in on relevance showing up here. then the phone number
200:50 of the lead, the email of the lead, etc. And here it's specifying how the AI
200:54 agent should be sending inputs into that. So that's probably the most
200:56 difficult technical part of this whole video, but I did want to give you a bit
200:59 of context on how that kind of fits together. This is quite a complex
201:03 schema. Relevance puts it together in a little bit more complex way um by using
201:07 these uh these schemas for the inputs down here. But long story short, if we
201:11 then go to the add or button, we need to set up our authentication, which we can
201:15 do by coming back to relevance and copying this. We go back to here, paste
201:22 this in. We go custom orth and we go authorize a orization with a
201:30 zed. Oh, I need to create the tool. Sorry. So, we can just click create
201:34 tool. So, the tool has been created successfully. And there we go. We have
201:38 both of the tools added in because we did them both in one bundle on
201:42 relevance. And then if we go edit off, we can then put in API keys for both of
202:05 one. And there we go. Now we have our knowledge set up, our two tools set up,
202:09 and you can see that we're pretty darn close to completing this build. We have
202:13 all of these three done. Might as well make them green for the sake of it. And
202:17 now the only thing left to do is to write a prompt that connects this all
202:20 together. And that's really the glue that holds it together. My go-to method
202:23 of rapidly creating prompts for AI agents is using a relevance tool. Um,
202:30 perfect that I've created, and I I said I'd give this to you guys for free as
202:33 well. That's going to be included in the resources. But if I go to use
202:38 here, it's a prompt writer that includes all of the information from how we do
202:41 prompting at Morningside, which is based on research and includes all the key
202:45 things like RO, task, specifics, context, um, explaining how to use the
202:48 tools that it's been provided as well. So, I'm just going to fill this out
202:50 quickly here and then get a prompt. And you guys can steal my prompt from the
202:54 resources or you can use this as well to create your own. But, it's a pretty good
202:57 exercise because you can see here we Um, and what I like to do here is, so
203:07 this is just a quick rundown of what the agent does, where it is being deployed,
203:20 why. So you can pause the video and look at that there. But just a bit of context
203:23 on what the agent does, where it is being deployed, and why. Conversions
203:45 contain. Then we get to the tools say and then we have the other tool
204:09 And then for the ideal input and output examples, I'm just going to say none to
204:20 assistant. And so just like that, in maybe a few minutes, I've typed in all
204:23 of this information about the agent and what it does. And now I can just click
204:27 run tool here. And it's going to take all of this information, run it through
204:31 the prompt that I've written that bakes in the best prompting practices for AI
204:34 agents from my experience and from the projects that we do at Morningside AI
204:37 and also all the research that we've used to make those prompting practices.
204:41 And it's going to spit us out an AI agent prompt that we can throw straight
204:44 into Agent and it'll just glue everything that we've done together,
204:47 tell it who it is and what it's trying to do, tell it how it's supposed to use
204:50 the knowledge base, and tell it how and when it's supposed to use those tools in
204:52 order to reach its objective of capturing those leads for us via
204:56 WhatsApp. And there we go. So, if we scroll down, we can see it's spit out
204:59 this entire prompt. I'm going to change it to the raw text so we get all this
205:04 markdown formatting included. We can view all here. I'm going to copy it all
205:10 and we're going to take it over to the here and paste this in. And there we go.
205:14 Act as corner cleaning WhatsApp support and lead generation agent. Engage with
205:17 potential customers on WhatsApp to provide potential information about our
205:21 cleaning services. Answer FAQs. answer off instant quotes ra when pricing inquiries arise use the
205:28 instant quote generator tool tools you have this tool and this tool examples
205:33 I'm just going to cut that out for now and then notes ra so that should be all
205:37 good we can start to give this a spin here I am going to zoom out a bit right
205:39 all right so I'm just going to publish this and make sure that everything is
205:44 baked in the second and now we can chat to it here um hey how's it
205:51 going actually slide this across Um, I want to know where you guys are
205:57 located. There you go. Connor's cleaning is located at XYZ. So, it's obviously
206:00 using the knowledge base correctly and say, "What what services do you
206:06 provide?" And we're not asking about quote or it might try to do it at the
206:13 end. Yes. So, see, it's asking if you have any specific requirements, need a
206:16 quote, just let me know. Yeah, sure. I'd like like a quote. Boom. I need the
206:21 property type and square footage. It's a house and it's 1,00 square ft. So now
206:26 the agent is trying to trigger that tool by taking the house and taking the 1,00
206:30 and then putting them into the relevance tool based off what the schema has told
206:33 it how to use the API. It's going to go grab that from relevance, send it back
206:37 to us and there we go. Here are the quotes for you. Ra, if you're interested
206:41 in any specific further service and need assistant, just let me know. I can also
206:44 help you with booking. Now here I would probably change the prompt and make it a
206:48 bit more forceful and say send me like like let's go to the next step right
206:51 now. But for now it's good enough. Um we can say uh sure I'd like to book a deep
206:58 clean. Now it should ask me for my lead information. Okay. Huge
207:04 Jackman is the name is the phone and huge Jack Jackman is email.
207:12 We should be able to see if we go back to our handy dandy air table
207:31 Oh, bang. And huge Jackman is in the CRM here. It does say that it's booked. I
207:34 would play around with the prompt a little bit more to be like, hey, look,
207:37 this is just setting up the next step for someone to call them and book in the
207:40 service. But you can also do appointment setting through agentive as well. Again,
207:42 like I said, it's a little bit more advanced than what we want to do here.
207:47 But as you can see, this is a very different way of approaching building
207:50 agents because you tell it, you basically provide all of the ingredients
207:52 and you use that kind of chef's approach. The knowledge and the tools
207:55 and you connect it all up and you make sure the tools have well described
207:58 schemas so they know how to use it and they know when to trigger them. The
208:01 knowledge base has been included in the prompt and also the tools as well have
208:04 been included in the prompt um telling it how and when to use it. It's really a
208:07 much faster way of building agents from the highle prompting and then people are
208:11 just asking and having sort of a free flowing conversation with it. Okay. And
208:14 just quickly before we go to the step of putting it onto WhatsApp which won't
208:17 take long at all. I do want to show you how you can debug and when you're
208:20 working in agentive um it's helpful to know when tools are being triggered and
208:24 why. So for example, if we go into the transcripts here and we look at this big
208:27 transcript here with 14 messages that we just had. Hey, how's it going? Ra. We
208:33 can see here it's using the tools and we can hover over it and we can see it's
208:36 calling the tool with the URL. It's a post method and we can see the data
208:41 here. I'll just zoom in on that. The property type and the square footage
208:44 that was sent away to relevance are here. So if you're having issues with
208:46 your tools or it's giving weird responses, um you can either come in
208:49 here to the transcripts after the fact. So say maybe this is on WhatsApp um and
208:54 something's going wrong or customers are getting upset. You can come into the
208:57 transcripts here and pick through and see what's going wrong with the tools.
209:00 And just like down here as well, the lead capture, we can see the name,
209:03 email, and phone were all put into this request and sent away to relevance AI.
209:07 And then onto Air Table as a second step. Um, and you can also see the
209:10 output as well. So the output of the tool is all in here. It's basically just
209:14 giving us a confirmation back from Air Table that, yep, everything went well.
209:19 And up here, you can see the output as uh the response with the deep
209:23 cleaning estimates and stuff like that. You can see it a lot more easily if we
209:32 Okay. And if you give this a second once it's finished generating agentive will
209:36 then pop up this show usage and bang there in the editor here. You can then
209:39 debug. Okay. How many tokens are being used? How much is this costing? What was
209:43 the model? Um etc. And then you can see the tools input here. Apartment 500 ft
209:47 etc. And the output as well. So it's really easy to debug those tools while
209:50 you're in Agent. Let's make sure that we've published this. I'm going to publish it again. In
209:55 Agent, we do have version history. So, if you do publish it and you want to
209:58 roll back or look at how you had it set up previously, you can now see that I've
210:02 got two versions, V1 here, and I just took away this little full stop here and
210:06 you can see that that's I've changed the prompt. So you can update it over time.
210:09 You can make edits within Agentive here and test test test. And then when you're
210:12 ready to push that to production and basically if we had this on a WhatsApp
210:16 agent and say I published this, connected to WhatsApp and it was working
210:19 and then I looked through the the transcripts and something wasn't quite
210:22 how I liked it, I could come in here and make edits and then test test and then
210:26 when I was ready to publish it, I click publish and then it's going to push
210:28 those live to the agent. So you're not going to mess things up by playing
210:31 around with things on here. So the final step is of course to deploy it to
210:38 Go to the deploy tab here. You can then click connect WhatsApp. I'm going to click
210:45 continue, get started. You will of course need a Facebook business manager
210:48 to set up this integration fully. That's free with every Facebook account. So, if
210:51 you haven't got one already, I'll leave a link in the description so that you
210:54 can set it up. Takes a few clicks. Then you will see this page here. And you can
210:56 select the business manager you've created. In this case, I'll be using
210:59 this testing one. And then you'll be able to set up a new WhatsApp business
211:02 account, which I can click here. I'll go next. set up a business account
211:09 name. And then this is the display name for the business. And we're going to
211:14 call this a a retail business. Now, you need to provide the phone number that
211:17 you want to connect your agent to. Um, unfortunately, you can't have your own
211:20 personal WhatsApp number and also have a business account running through it. So,
211:23 you need to either buy another SIM card or borrow a friend's number who doesn't
211:26 have WhatsApp, etc. In this case, I'll be using a spare number that I have.
211:29 Then, they're going to send you a verification code to your number, which
211:32 you have to enter in. And then you should see the screen once you've
211:35 successfully passed that verification. So when we continue, it's verifying our
211:39 information for a second. And now our agent is connected to that phone number
211:42 and we're ready to give it a test. So if we go finish here, there's one more
211:45 thing that we need to do on Agentive, which is to click this. Yours may say
211:48 not registered. Don't worry, you can just click this check box here and click
211:53 confirm. Give it a second to connect. Now we've successfully connected our
211:56 agent to that WhatsApp number. Now thing here is this interval. If you're not
211:59 sure what the interval is, you can read this tool tip here. And if you're done
212:02 with the deployment and you want to remove it from that number, you can
212:04 always come back and click deactivate deployment here. But all that's left to
212:07 do now is to test our functionality. Right. So I have it connected to my
212:10 phone here. So I'm just going to show you a little bit of a on screen here of
212:13 me creating this contact and having a message with it. So the number that I
212:17 set up, I can create a new contact and Hey, and you can see on screen here it
212:26 says this is the business using a secure service from Meta. So, this means this
212:29 is a business account um as we've connected it through our WhatsApp uh
212:33 business profile that we set up before. And there we go. We get a
212:36 message back. Hello, thank you for sharing your information. How can I
212:39 assist you today? Um if you have any questions about cleaning service or need
212:42 a quote, feel free to let me know. So, I can say um yes, a quote. Let's ask a
212:47 question to the knowledge base. Where based? There we go. We are based in the
212:55 greater Boston area. It's giving me the uh the correct location there. So, we
212:58 can go for the lead capture now. So, if quote. So, property type uh it's a house
213:12 that is ft. There we go. We're getting the uh estimations and our quote back. Um, it's
213:24 asking if we're interested in any of these services. I'd say yes, I'd like
213:33 please. Now, it should ask me for my contact information. There we go. So, Liam, I
213:41 mean, Liam at mail. Cool. And then we should see it appear over here on our Smith Solar
213:49 CRM. And boom, there it is. So, we've got everything done. That is just one
213:52 run through of using this WhatsApp agent. But as you can see, uh the the
213:56 messages don't come back instantly. So it it feels like it is like it's
213:59 actually could be a real human there applying and it's giving just clear
214:03 information right through WhatsApp. Imagine you are reaching out to maybe
214:06 book an accommodation or you're reaching out to a a cleaning service like this or
214:09 you're reaching out to any kind of business and you want some real
214:11 information directly from what feels like a person. And then you also have
214:15 the functions of getting a real quote of I mean a great use case for this kind of
214:19 thing is like barbers. Like I say, if you you message a barber on WhatsApp,
214:22 maybe you're in in Europe somewhere, you're in South America or you're in
214:25 Central America or and and you want to go to a barber and this is a common
214:28 issue that I've run into when I'm traveling. It's like, I want to message
214:32 this barber, but I might not speak the language that well. And then if you
214:34 message them in English, it will be able to handle that in in English as well as
214:38 in Spanish or in Portuguese or wherever you are. So, this kind of functionality
214:40 built through WhatsApp is a really really great use case for you guys to
214:43 pick up, which is why I wanted to teach you guys it. And we can also go back to
214:47 agentive here. And if we go to our transcripts for this agent, we can see
214:51 the one for today is here. So 12 messages. You can go through the entire
214:54 transcript and you can see it's calling the quote tool here. We see all the
214:57 information that went in and out of it. And then we see the air table lead
215:00 capture information as well. Input and output. So that is how you use a genty
215:04 my software for building these WhatsApp based and also other deployments as
215:07 well. So if we go to studio and we go to deploy, we have Instagram as well. So
215:10 via our mini chat template. You can hook into Instagram and do appointment
215:13 settings and things on Instagram. You can go through Messenger if you want to
215:17 run some Facebook lead ads to Messenger through voice flow as well. Telegram,
215:20 Discord, we have integrations with everything you need as well. So that's
215:23 the end of this build. I hope you enjoyed and uh this is a super handy use
215:27 case um and and deployment really for agents. So now that you understand how
3:35:29 The Real Opportunity
215:36 AI agents works and can build them for yourself, let's talk about the most
215:39 important part of this, which is actually making money with these skills.
215:43 But first, we need to destroy a huge misconception and that you don't need to
215:47 build the next chat GBT or create some revolutionary AI startup in order to
215:51 make money in the AI space. The real opportunity is much much simpler. It's
215:55 just helping businesses to understand and implement AI. This is how I
215:59 monetized my AI agent skills and it has been the most explosive growth I've ever
216:03 experienced in my career. And the good news is, if you've made it this far in
216:06 the video, you are so much closer to being able to tap into this starving
216:10 market for AI services than you think. But don't take my word for it. I'm just
216:13 some guy on the internet after all. Maybe you should listen to some of the
216:15 world's most famous businessmen saying that this is the opportunity to get into
216:19 right now. If I was 25 years old today in 2024, what would I do? What's a good
216:24 sector to get involved in? What business would I get involved in? I think
216:28 everything is looking at AI now in a different way. And I think AI growth is
216:32 going to be exponential. So, anything to do with AI now, what could that be? In
216:36 the simplest form is helping people use the technology. there's going to be a
216:40 massive amount of people wanting to use it that don't know how to and they're
216:44 willing to pay to solve that pain point. So, is that consulting? Not really. It's
216:50 implementation and execution. And so, helping a business do that transfer into
216:54 a world where they're controlling their data and getting information from it.
216:58 Now, the majority of businesses in America, for example, are between 5 and
217:02 500 employees. So, they're small businesses. They create 62% of the jobs.
217:07 They want to use AI. you should help them solve for that and they'll pay you.
217:11 Even another shark, Mark Cuban, is saying the exact same thing that the
217:14 biggest opportunity right now is helping these small to mediumsiz businesses who
217:19 don't understand AI yet, but desperately need it to keep up. And they're
217:22 absolutely right. If we look at the data, it's pretty obvious. According to
217:26 recent data, there's 1.7 million businesses in the US alone that are
217:30 making between $500,000 and $10 million per year. These are small businesses,
217:35 which, as Kevin Oer says, make up 62% of the jobs in the USA. They create 62% of
217:40 the jobs. They want to use AI. You should help them solve for that and
217:43 they'll pay you. These businesses know they need AI to stay competitive, but
217:46 they don't have the time to learn it themselves. And there's basically no one
217:50 there to help them. All of the big consulting firms are looking at other
217:53 big businesses and just leaving these smaller businesses completely ignored,
217:57 but they still make lots of money and they still have a lot of money to invest
218:00 in these kinds of services. Basically, all small businesses are starving for
218:03 some kind of AI services, either education services to help them
218:06 understand what AI is in the first place and why they might need it. There's the
218:10 huge need for consulting services where you help them to identify where AI can
218:13 help with them most in their particular business. And of course, there's
218:16 implementation services where you help them to build and maintain the AI
218:20 systems like the AI agents we've just built. And right now, based on the data
218:23 collected in my community, and we are the largest AI business community on the
218:26 planet right now, for every person or agency that is currently offering AI
218:30 services, there are over 1,100 businesses in the USA alone that need
218:35 help. So, that's a 1 to 1,100 ratio, which means this is a completely
218:38 untapped market. And that's where people like you and I come in, helping these
218:41 hardworking small business owners to understand AI and implement it so that
218:45 they have a chance to keep up. And that's really what drives me and the
218:48 team at Morningside because our company vision is a world where the benefits of
218:52 generative AI are distributed as fairly as possible and they make it to people
218:56 like me and you and the small business owners rather than just all going to
218:58 these giants at the top. And this whole concept of selling services around an
219:02 emerging technology is nothing new. And we saw the exact same pattern when the
219:05 internet first came out. Companies that helped businesses to adapt to the web
219:09 and sort of get online made fortunes. You know, agency.com, Razerfish, etc.
219:13 And I spotted this opportunity within the AI space in 2023 when it wasn't
219:17 anywhere near as clear as it was now. No one really knew how to make money out of
219:20 this stuff. And then I started Morning Side AI. And since then, we've generated
219:24 over $5 million in selling these kinds of AI services of education, consulting,
219:28 and implementation. And we're literally still only just getting started. And the
219:31 best part out of all of this is that as we've proved in this video already, you
219:35 don't need to be a technical genius to understand AI and even to build AI
219:39 agents. You just need to be one step ahead of the businesses that you're
219:42 going to be helping. So, let me show you the three specific ways that you can
3:39:47 Three Ways to Win
219:47 start making money with your AI agent skills. So, as I said, there's basically
219:51 three types of services that you can provide to businesses in order to
219:54 monetize your skills. Firstly, there's education, and this is teaching
219:58 businesses about AI, running workshops, and doing presentations, training their
220:01 teams, and creating courses for them to watch. Businesses are desperate for
220:05 someone who can explain what this stuff is in simple terms and more importantly
220:08 what it can do for them. After watching this video and probably my other huge
220:12 video that I did on AI agents, which I'll link down below, um you will know
220:16 more than enough to start educating businesses on AI and AI agents. Secondly
220:20 is consulting. And this is where you analyze a business's operations and you
220:23 show them where AI can help them save time or make more money. You're
220:27 essentially being their AI strategist. For example, you could go into a
220:30 business and then recommend something like the sales co-pilot system that we
220:34 just made in order to help their struggling sales department. And third
220:37 is implementation. So this is where you actually build and deploy AI solutions
220:41 for businesses. Or better yet, like my agency, you can do all three of these,
220:45 but it did take us 2 years to get here. So there is really no rush. You just
220:48 pick where you want to enter and work your way up to doing more and more if it
220:51 makes sense. Believe it or not, there are people with only a few months
220:54 experience in the AI space selling all of these right now. And the demand from
220:58 businesses is increasing insanely fast right now. I we're seeing this at
221:00 Morningside. Just so many more businesses reaching out. But here's the
221:04 thing. You have one small problem, and that's that you don't quite know enough
221:08 to start moving on this. You are close, but you're not quite there. The way to
221:11 make money in the AI space or with any services really is to create a knowledge
221:15 gap between yourself and the people that you're helping. Your knowledge gap is
221:18 your money maker, and businesses will pay you in proportion to how much more
221:22 you know about AI agents and their business applications than they do. Now,
221:26 while this video has taught you a lot, your knowledge gap is still small. But
3:41:30 Extending Your Knowledge Gap
221:30 we can fix that. So, let me break down exactly what you need to do next in order to extend
221:35 your knowledge gap to the point where you can start making money. We can call
221:39 this video as step one. So, as long as you've taken notes and followed all the
221:42 tutorials and built the agents alongside me, you're already ahead of most people
221:46 who have no idea about what agents are, how they work, or how to build them. So,
221:49 it's a big step forward with this video. But step two is building even more
221:54 experience building AI agents. So you are more familiar with the platforms and
221:57 better understand the different ways that they can be used to deliver
222:01 different kinds of AI agent use cases or even just AI tools in general. I've only
222:04 really given you a taster here, but I tried to make it as as diverse as
222:07 possible as you could probably tell. In order to do this second step of
222:10 extending your knowledge gap further and building more experience, you can go to
222:14 my free course on school where you'll be able to build another 5 to 10 agents
222:17 following the tutorials that are in there for you. So the link to join my
222:20 free school will be in the description. So, if you blast through all those
222:22 tutorials in there, this is going to further expand your knowledge gap. And
222:26 remember that the more that you know compared to the businesses that you're
222:28 trying to help, the more they're going to pay you. So, step two is building a
222:32 few more agents out and getting a bit more experience on the tools, seeing
222:35 different use cases, etc. And once you've done that, you'll have what I
222:38 call foundational knowledge. So, you understand the core AI concepts that
222:42 we've been through in this video. You can build basic solutions on these
222:45 platforms. You know what's possible for businesses with agents right now. And
222:49 then comes the big decision. Do you want to go deeper technically on this
222:52 building side of getting your hands dirty or do you want to start monetizing
222:56 what you already know? As we've covered, the building and implementing of the AI
222:59 systems is only one of the services that you can sell. So naturally, the
223:03 technical skills needed in order to make money in the implementation services.
223:07 Actually building these systems and businesses is much more greater than
223:11 just having a foundation. But with a good foundation, you're basically ready
223:14 to start having a crack at the other two services of AI education and AI
223:18 consulting. So, the decision of what to do next comes down to really knowing who
223:21 you are and what you are really interested in. And this sounds all woo
223:24 woo and like, oh, you got to know yourself and stuff, but this is I mean
223:29 it very very seriously in that if I use myself an example, I've always loved
223:32 making things, right? I used to build block houses with kids. I used to like
223:36 brew beer with my grandpa. I've always loved tinkering with engines. So when I
223:39 hit this foundational level that you guys will be at after completing those
223:42 extra builds in the free course, I kind of naturally just dove deeper into the
223:45 technical side into building more stuff. I I kept building more and more complex
223:49 AI systems and building upon those skills that I've I've built already,
223:52 which led me ultimately to starting Morningside AI where our first service
223:56 was building AI solutions and systems for clients. But here's the thing, of
223:59 course, a lot of people aren't like me. They don't get as much of a buzz out of
224:03 building things. many of you are going to be much better or enjoy more the
224:06 teaching aspect or working with people and doing the consulting aspect rather
224:10 than building stuff. So in these cases using the foundational knowledge that
224:13 you're going to build up to sell AI education to businesses or AI consulting
224:17 makes a lot more sense. Goes back to the whole Einstein thing about like judging
224:21 a fish on its ability to climb a tree. If to you the building is like a tree
224:23 and you feel like a fish and it's not a really good fit, then there's better
224:26 stuff that you can do and you can find a way to make money in the AI space that
224:30 leans more into your strengths. like in the case of a fish would be swimming,
224:33 right? So, by being honest with yourself and saying, "Hey, look, that's not
224:36 really me. Yeah, sure, I did get it done. I know how that works now, but I
224:39 don't feel any kind of attraction to doing more of that." While you may see
224:42 that as a negative and saying like, "Oh, I don't have what it takes." It's
224:45 actually can be very empowering if you say, "Bang, I'm stopping it here. I'm
224:48 stopping the learning. I'm stopping the procrastination. Now, I'm going straight
224:51 into actually monetizing." It's basically putting a stop on when you do
224:55 this learning big long phase and saying, "No, action starts now. I'm not I'm
224:59 never going to get there, but with this base, I can do a lot and I'm going to
225:02 start taking action with it and making money with it today. So, this
225:05 self-reflection is really what prevents you from getting stuck in an endless
225:08 learning phase of procrastination when you could be out there making money. So,
225:11 in summary, the two routes you have and the two options you have from here are
225:15 if you love building and you kind of naturally feel like you want to learn
225:19 more like like myself when I was at your stage, then just keep going. go and
225:22 watch the free course tutorials on my school and then start going and building
225:25 your own projects and ones for friends and family and whatever you you sort of
225:29 pulled towards naturally and within two to three months you'll have enough
225:32 skills and experience to actually start selling implementation properly. But on
225:35 the other hand, if you haven't fallen in love with the building aspect, then it's
225:39 probably best that you just go go into the free course, smash out the rest of
225:42 those tutorials and finish your foundation and then just get started on
225:45 monetizing your skills either through selling AI education or through selling
3:45:49 Getting Your First Clients
225:50 AI consulting. So once you're clear on what kind of AI services you want to sell, getting your
225:55 first few clients is actually pretty straightforward. People try to over
225:58 complicate it, but there's really just two main ways that I'd recommend you do
226:01 this based off all the success I've seen over thousand thousands of people across
226:04 my free and paid communities. The first and by far the easiest method is through
226:07 your warm connections or warm contacts. This means reaching out to people that
226:09 you already have some kind of relationship with, whether it's friends
226:13 or family or kind of acquaintances or even friends of friends that you've met
226:16 once kind of thing. All of these people count as warm connections. So instead of
226:19 trying to convince complete strangers to trust you with your business, you can
226:22 start with people that you already know or have some previous relationship with
226:25 and therefore have an increased level of trust with you through your
226:28 relationship. And I've covered this many, many times on the channel before.
226:31 So on the school post for this video, I will add in my complete guides for warm
226:35 outreach, including resources directly from my AAA accelerator program. The
226:38 second way is using what I call the community content flywheel. So this is
226:42 how you can build long-term momentum beyond just warm outreach. So here's how
226:46 it works. You join the free school community. you start making content
226:48 about what you're learning at each stage. This could be through YouTube
226:51 tutorials, which I mean that worked for me. LinkedIn post is another one um or
226:55 whatever platform you really prefer to create content on. But here's the key.
226:59 You share that content back into the community. So with over 120,000 members,
227:05 by posting it into the community, you get an instant audience and people who
227:08 are really interested in the stuff that you're talking about. So, a perfect
227:11 example of this is a guy called Rory Ridges, a a young guy from the UK who
227:14 joined my free community and basically followed this exact process that I've
227:17 told you in this video so far. So, he took my free course, he learned all the
227:21 basics, built his foundation, then he started posting simple tutorials on
227:25 relevance AI, which you've used in this video already, and he literally just
227:28 started sharing what he'd learned from my videos and making other videos about
227:31 it. And at the start, he was literally just sharing what he'd learned from my
227:34 videos. So, he'd watch a video, then go and kind of make his own video on the
227:37 same sort of topic. And every time he made a tutorial, he'd then share it into
227:41 the community and the community would watch it. They'd give him feedback, go
227:44 and subscribe to his channel. This not only helped him grow faster on YouTube,
227:47 but it also started to position himself as an expert in the community. And he's
227:50 also building his authority in the AI agency space by getting more momentum on
227:54 YouTube. Now, his YouTube channel brings him in enough leads to support his
227:57 growing agency. And I've just seen him recently in the community saying he's
228:00 hiring. So, that's usually a bloody good sign that he's making some good money
228:02 off the back of it. He basically started the same flywheel that took me from zero
228:07 to where I am now. Over $5 million in revenue generated across all my
228:11 businesses and $450,000 plus subscribers in just two years. And so the community
228:15 gives you an audience and the content gives you credibility and together this
228:18 method brings you clients. In the resources for this video on school, I
228:22 will leave links to my complete guide for creating content to generate leads
228:25 just like Rory and I have done successfully. And of course, I'll
228:28 include a link to Rory's channel in the resources in the school community. Now,
228:30 the really important thing to notice with both of these methods we've just
228:34 talked about, the warm outreach and the uh community content flywheel is that
228:37 both of these methods start with giving value first. Whether it's helping your
228:41 warm connections to understand AI or sharing your knowledge through content,
5:39 What Are AI Agents?
5:39 let's get stuck into it. All right, so step one in building AI agents is knowing what the hell an
5:45 agent actually is. So, 2 years ago, when I first started learning about AI
5:49 agents, I had no idea what they actually were. The term AI agent gets thrown
5:52 around a lot in almost like everywhere these days. You got AI agents this, AI
5:57 agents that. But what actually is an AI agent? Well, the clearest definition
5:59 that I found that helps beginners to really wrap their head around what they
6:03 are is this. An AI agent is a digital worker that can understand instructions
6:08 and take actions in order to complete tasks. So, in a very simple way, just
6:12 like businesses have employees who handle different tasks, an AI agent is
6:17 like having a digital employee. But the cool thing is that you can build them
6:20 and you can make them do whatever you want. You're like literally building an
6:23 employee that you can put to work to do things for you. And of course, they cost
6:27 much less to run than a human and they don't need sick days and they don't
6:31 start beef with Mike over in the sales department because of his comment at the
6:34 coffee machine. So, I'm sure you can see the appeal of this kind of digital work
6:38 and AI agents to businesses who are looking to adopt them. In order to really understand why these
6:40 Chatbot or Agent?
6:45 AI agents are such a big deal, we need to look at where we are coming from. So
6:48 most of you have probably encountered those chat bots on websites before. You
6:52 know those little little chat widgets that pop up saying like, "Hey, how can I
6:55 help you?" So these kinds of chat bots are pretty basic, right? They a lot of
6:58 the time they're useless and they're they're kind of like a waiter who can
7:02 only really recite the menu but can't actually take your order or or bring
7:05 your food. They can't do anything. They just respond with some kind of
7:08 pre-written answers. Well, nowadays it's a simple AI generated answer. But AI
7:13 agents are different, right? So, here's an example. If you ask a regular chatbot
7:17 about booking an appointment, it might say, "Oh, our business hours are 9 to5.
7:21 Please call to book." And that's it. They just give you some information
7:25 back. But with an AI agent, it could actually go and check the calendar, find
7:29 some available slots, go back and forth with the person that they're chatting to
7:32 in order to book an appointment, send you a confirmation email, then update
7:36 the business's scheduling system and CRM automatically in seconds. This ability
7:41 to take action is what makes agents so powerful. They're not just fancy chat
7:44 bots. They're actually digital workers who can search through databases, update
7:49 spreadsheets, send emails, book appointments, generate hold documents,
7:54 and much much more. And so building and deploying an AI agent is a bit like
7:58 hiring a new employee because when you bring someone into a business, you need
8:02 to firstly explain their roles and the responsibilities to them. You need to
8:06 give them access to your system so they can use them. And you need to trust them
8:10 to handle those tasks independently. And now when we are building agents, as we
8:14 see later, it's exactly the same, except these agents are going to be working
8:17 24/7. They're never going to get tired. They can be duplicated and modified
8:21 instantly. And they cost a fraction of what a human employee does. And this is
8:25 exactly why understanding how to build and sell AI agents is becoming such a
8:29 crucial and valuable skill these days. Because whether you're an entrepreneur
8:33 looking to scale your business or you're an employee wanting to become
8:36 irreplaceable and and make more money at work, knowing how to create and deploy
8:41 these digital workers is like the biggest cheat code in the whole world
8:44 Anatomy of an AI Agent
8:44 right now. Now that you understand what AI agents actually are, let's look under
8:50 the hood and see how they actually work. Just like humans need a brain, memory,
8:55 and tools in order to do their job, AI agents need specific components in order
8:59 to function correctly. An AI agent needs five key parts in order to work.
9:03 Firstly, every AI agent needs a brain. In the AI world, we call this a large
9:07 language model or an LLM for short. And you've probably heard of some of these.
9:11 You've got GPT from OpenAI, Claude from Anthropic, Gemini from Google, etc. You
9:15 can think of the LLM as having a super smart intern who can understand your
9:19 instructions in plain English, and then figure out how to get things done from
9:22 those instructions. So, without this brain, all of the other parts would be
9:25 useless, right? It's like having a whole desk full of office supplies but having
9:28 no one sitting there in order to use them. Secondly, the brain needs
9:32 instructions on how to behave. And this is prompting. So writing a prompt for an
9:36 agent is how you program a lot of the behavior of it rather than having to
9:39 code it manually. And this is really what makes building AI agents so much
9:43 more accessible to non-coders as the way of actually programming the
9:46 functionality and how they work is done through clearly written instructions
9:49 rather than having to actually code it. Thirdly, agents need memory. Imagine
9:53 trying to have a conversation with someone who forgets everything you said
9:56 30 seconds ago, right? So, memory is really important because it allows your
10:00 agent to remember what you talked about just a few messages ago, keep track of
10:04 the tasks that it's been working on, build on previous conversations, and
10:08 even in more advanced ones, it can learn from your past interactions. And the
10:11 good news about memory is that most AI agent platforms completely handle this
10:14 memory component automatically. So, you don't need to worry too much about it.
10:17 But just know that it is an important part of a functioning AI agent. The
10:20 fourth component of an agent, and this one is optional, but it is external
10:25 knowledge. AI models like GPT and Gemini are pre-trained on a huge amount of
10:29 data, but that data is basically cut off at a certain point, eg 2024. It's kind
10:34 of like having a new employee who only really knows what they learned in
10:37 school. But just like you can train an employee like that with your company's
10:41 specific materials, you can also give an AI agent additional knowledge on top of
10:45 the information it was trained on through providing things like PDFs of
10:48 your company documents, spreadsheets with product information, customer
10:52 service transcript, or basically any other textbased information. Without
10:56 this added knowledge, agents will be limited to general information and
10:59 couldn't handle specific business tasks. But as I said, knowledge is optional and
11:02 you will only need it in some builds. Finally, and this is the most important
11:07 part, we have tools. So tools are what transform an AI agent from just being
11:11 able to chat to being able to actually get things done. So you can think of
11:14 tools like giving your digital employee access to different softwares. Just like
11:18 you might give a new hire access to your email or your calendar or your CRM
11:22 system, you can give an AI agent access to digital tools that let it take
11:26 actions when needed. These tools let your agent do things like checking
11:29 real-time data, updating databases, sending messages and notifications,
11:32 creating documents, all the stuff we went over just before and much, much
11:35 more. The really powerful part, which we're going to cover later, is when
11:38 agents use multiple tools together in order to solve complex problems, just
11:42 like us humans would use multiple different websites and softwares when
11:46 doing our tasks. Now, let me show you how all of these parts work together in
11:50 a real example. So, say you want an agent to handle customer support. When
11:54 the agent is sent a message, the brain immediately understands the prompt that
11:57 it has been given and also understands what the customer is asking. It checks
12:01 its recent memory before replying each time to understand the full context of
12:04 their conversation. And if the brand detects that the customer wants a
12:08 specific question answered from the knowledge base, it will use its external
12:12 knowledge in order to deliver the right information to them. And finally, it may
12:16 use tools to update a customer's account or to process a refund whenever required
12:20 during the conversation. So all of these things are happening in seconds as the
12:23 conversation is going on. Which is why AI agents are such a game changer. They
12:27 can combine all of these components in order to create a fully capable digital
12:31 worker that very very closely replicates how humans work. Now that you know the anatomy of
12:38 an agent and the five parts of it, a more practical framework for
12:41 understanding how we actually plan and build AI agents is what I call the three
12:45 ingredients. Basically, you only have three elements to plan when creating an
12:49 AI agent, which when mixed in various ways can create millions and different
12:53 types of agents for different use cases. This is because the AI model or brain
12:57 can be easily swapped in and out and isn't really a major factor in the
13:01 performance of the agent as any of the top models that you pick from any of the
13:04 different providers at any given time, they're all pretty good. And also, the
13:07 recent chat memory is handled by default in almost all cases when you're building
13:09 on these platforms that you're going to see later. What this leaves us with is
13:12 what really matters when building and planning AI agents. Firstly, the
13:15 knowledge, the external data that you want the agent to be able to use when
13:19 answering. Secondly, the tools, the different actions that you want the
13:22 agent to be able to take, eg saving the contact info to the CRM or getting some
13:27 live data on stocks or sending an email. And then finally, prompting, which is
13:31 the glue that ties everything together and determines how the agent behaves.
13:35 So, write these down. While the agent has five components, the brain, the
13:38 prompt, the memory, the knowledge, and tools, your main focus as an AI agent
13:42 builder is in the three ingredients of prompting, knowledge, and tools. In the
13:45 next chapter, we'll be looking at how you actually build an agent using the
13:48 different combination of these three ingredients. But first, we need to dive
13:52 deeper into the keystone of understanding how to build your own
13:55 valuable digital workers. And it all comes down to tools. Now, we need to dig a lot deeper
14:02 on tools as they are by far the most powerful part of AI. agents. But in
14:06 order to understand deeply and be able to build powerful agents with them, we
14:10 need to take a few steps back and actually cover the basics of how
14:14 software and the web and internet as a whole works. Now, this is as techy as
14:17 it's going to get in this video, but I promise once you understand this, it's
14:20 so important and it's literally like having a superpower. So, please stick
14:24 with me through this. So, remember how we said that tools are what allow agents
14:28 to take an action to actually do things rather than just chat? Well, the way
14:33 agents use tools and do work online is just how we do it as well, but with one
14:37 key difference. Instead of clicking buttons and typing into forms, agents
14:42 use what we call APIs. And every time you use the internet, you're actually
14:46 making dozens of requests to APIs as well and getting responses back, but you
14:49 just don't realize it. So, let me show you what I mean. So, when you click on
14:53 this video, here's what actually happened. Firstly, your browser sent a
14:59 request to YouTube servers saying, "Hey, I want to watch this video." And then
15:02 YouTube servers sent back all of the data needed. And thirdly, your browser
15:07 unpacked that data and started playing the video on your screen. So this
15:10 request and response pattern happens with almost everything that you do
15:13 online. When you open up Instagram, you are requesting your feed from Instagram
15:17 service. When you send a tweet, you are sending your data through Twitter's
15:20 service. And when you check your email, you are requesting from Google the
15:24 latest messages in your inbox and they're sending it back and your browser
15:27 is loading it. Thankfully, we get pretty websites and apps that make it very easy
15:31 for us to do this and use software via APIs through a nice application. But
15:34 under the hood, it is still two computers talking back and forth,
15:38 requesting, sending, and displaying new information for us on our screen. These
15:41 request and response happen through what we call APIs, which are application
15:45 programming interfaces. So, you can think of APIs like waiters in a
15:48 restaurant. Basically, they're going to take your order or your request to the
15:52 kitchen, which are the servers of the business, and then they bring back your
15:56 food, which is the response. So, you have request and response, and you have
15:59 you as the client, and them as the server. There are two main types of
16:02 requests that you can make. Firstly, either a get request. This is basically
16:07 just like asking for information like checking the weather or looking up the
16:10 price or loading this video. You're requesting to get the information to do
16:14 something. Secondly, we have post requests, which is when you're sending
16:17 some kind of information like posting a tweet, sending an email, or uploading a
16:21 photo. So, go back and write both those down because we're going to be using
16:24 them extensively in the building section of this video. Now, here's where it gets
16:28 interesting. So, AI agents use these same APIs as their buttons to do things.
16:32 So, each tool an agent has access to use is essentially an API that it is able to
16:36 call. So, these kinds of tools come in two different flavors. We have pre-made
16:40 integrations like Google Calendar or Gmail where it kind of comes out of the
16:43 box ready for you to use and just plug straight into your agent. And then we
16:47 have custommade tools that we can build ourselves. So you can think of pre-made
16:50 integrations like buying a readymade meal where they've done a lot of the
16:54 hard work versus custom tools where we are like cooking from scratch. And both
16:58 work, but custom tools give you a lot more control. And this is a skill that
17:01 I'm going to be teaching you in the second chapter of this video. Okay. So now you got the basics.
17:07 Let's get clear on how a tool is actually made and what the key parts are
17:11 as you're going to be using them a lot. So, let's break this down using a simple
17:15 example of a text capitalization tool. It takes in some text and the outputs
17:19 the capitalized version of it. So, first to create a tool, we need a function. We
17:23 need something that does work. In this case, it's super simple, right? It needs
17:26 to take in text and it needs to make it uppercase. So, this can either be done
17:30 through a basic Python function or you can use an LLM to do this as well.
17:33 Basically, we need to build some way to capitalize the text that we give to this
17:37 function and actually do the do the work. Next, in order for the AI agent to
17:41 use this function, we need to wrap it in an API. So, we have the function and
17:44 then the API wraps around it. And this is essentially making that functionality
17:49 we created accessible over the internet via APIs. Without it, the function
17:53 cannot be used by our agent. And in order to use the API that we've just
17:56 created and use this function inside it, the API is going to expect the same sort
18:01 of inputs that the function needs. So the input of the text that we want to
18:04 capitalize and it's going to output the capitalized version. So this is very
18:08 important to remember function. It takes in the input of the uncized text does
18:12 work and outputs the capitalized version. We're basically then just
18:16 building an API around it so that we can put it on the internet and then we can
18:19 have an agent that knows how to call that API can send information into the
18:23 input go through the function and then get spit out and then our agent catches
18:27 it at the end. But the magic step and what has really caused the AI revolution
18:31 to kick off is that we can explain to our agent how to use this API just by
18:36 explaining how the API works in natural language. And this is where schemas come
18:42 in. A schema is like a one-page instruction manual on how to use an API
18:47 and therefore how to access the functionality inside that API. And when
18:50 an AI agent is given one of these schemas, it too can read that
18:54 instruction manual and determine things like what the tool does, what
18:58 information it needs as an input, like we talked about before, and what
19:02 information to expect as an output. Now, they may look scary, but they're
19:04 actually really, really easy to understand, and we're going to cover
19:06 them in the next chapter with this video. And the good part about it is
19:09 that these days, schemas are automatically created by many of these
19:12 no code platforms that you build agents on. But I'm teaching you this because it
19:16 still helps to know what they are doing and what that what's really happening
19:19 under the hood on these platforms. And there are still going to be times where
19:21 you may need to roll up your sleeves and do it yourself. The incredible part
19:24 about these schemas is that modern AI like chatpt can read these instructions
19:28 and perfectly understand not just how to use it and like okay I need an input and
19:32 then I expect an output but also when to use it. For example, let's say we had an
19:36 agent and we gave it that capitalization tool that we just talked about and then
19:39 we said can you please capitalize this text? Mary had a little lamb. The agent
19:43 would then read over the schemas that we provided it and then it would see that
19:46 there's a tool with a description saying this tool capitalizes text right in the
19:50 instructions for the capitalization tool. We would have said this thing is
19:54 for capitalizing text and it takes in some text and it gives you the
19:57 capitalized version. And so the agent will read that and see okay this looks
20:00 like based off the instruction they just gave this is the tool that they want to
20:03 use. And then it will check the requirements and see that the tool takes
20:07 in one input in string format which is just text which we have described as the
20:11 text to be capitalized. So it reads all this. He says okay it it needs one
20:15 input. It's in string format. So I know I need to give it some text and okay
20:18 what does this text do? It's the text that they want to capitalize. Great. So
20:22 now it knows it needs the input and it knows that this is where it's going to
20:24 send the text to be capitalized. Then now that it knows what it wants, it goes
20:28 back to our message and it intelligently extracts Mary had a little lamp. not,
20:32 hey, can you please capitalize Mary had a little lamb? It's smart enough to know
20:35 that we want that taken out. So, it will take that part, Mary had a little lamb,
20:39 out of our input, and then it sends that to the API where our capitalization
20:43 function does its thing. Then the API sends back the capitalized version plus
20:48 a bunch of other response data as well. Then the agent looks at your original
20:52 question, looks at this messy response it got back from the API, and then using
20:56 its brain, the LLM, it writes a natural language response answering your
20:59 question. It would say, "Here's your capitalized text colon Mary had a little
21:03 lamb in all caps." That may sound complicated. It may have gone over your
21:06 head. Please go back and just listen to it again. You really, really need to
21:10 understand this process of uh the message comes in, looks at the schema,
21:13 realizes, okay, it wants to use this tool. Okay. What do I need to do in
21:16 order to use this tool? Okay. Well, then I'm going to grab it out of the input.
21:19 I'm going to put it in here. And it can actually go back and forth. Say our
21:22 capitalization tool needed some other input. Say you needed to provide uh the
21:26 number of letters you wanted to be capitalized. It may see that this tool
21:30 needs two inputs and I've only been given one. So then it will go back and
21:34 ask me, hey, could you can you please tell me how many letters you want to be
21:37 capitalized and you will see this magic in the agents that we're going to build.
21:40 When the agent can ask you questions in order to help fulfill the needs of the
21:43 tool, you have this very intelligent system that really will blow you away
21:46 when you see it in action. And one thing many people miss about this process is
21:50 the agent actually gets back raw computer data from the API or what we
21:55 call JSON. But using the LLM, it can transform that into natural conversation
21:59 and answer your question in a very very uh clear and concise way. So it's
22:02 basically like having an employee who can read all this technical information
22:05 and then explain it to you in plain English, which is another part of why AI
22:09 agents are so powerful. And so when you understand this pattern that we've just
22:11 gone through, I promise you, you will never see the internet the same way
22:15 again. Every action online is just requests and responses. And therefore,
22:20 we can build our own tools and AI agents to automate all of it. So instead of you
22:24 manually searching the web, copying information, pasting it into
22:28 spreadsheets, sending emails, an AI agent can do it all automatically using
22:31 tools if you build it correctly. It's like having a digital employee who can
22:36 press all of these API buttons for you thousands of times faster than any human
22:39 could. And don't worry if this feels a little bit technical. In the next
22:43 chapter, uh I'm going to show you how to create your own tools like this from
22:46 scratch using platforms like Relevance AI, uh where you can build out powerful
22:49 tools without writing any code. and will really start to click into place once
22:52 you see the stuff in action in the building section. But before we get into
22:55 that, let me reveal the power of AI agents which is unleashed when they are
23:00 given multiple tools to work with. Now, obviously having an AI agent
23:04 that just capitalizes text isn't very useful. I get that. The real magic
23:08 happens when you give agents multiple tools and the ability to use them
23:11 together in order to achieve complex goals. So, do you remember our
23:15 definition? AI agents are workers that can understand instructions and take
23:19 actions to complete tasks. When you give an AI agent a task, it's going to try
23:23 its best to execute on it, but if it doesn't have the right tools on hand to
23:26 do the job, it's going to be useless. And so, the more tools that you can give
23:30 an agent, the more flexibility it has to solve problems just like a human would.
23:33 So, let me give you a real example from my own business, right? Say I build an
23:37 agent and give it the task. Find AI startups that have recently raised money
23:41 and put them in a spreadsheet and add a summary of each of the businesses in the
23:44 spreadsheet and then email me the link to the spreadsheet. When you give an AI
23:47 agent a task like this and provide it with multiple tools to use, it can break
23:51 down this problem just like a human would. For example, it might think first
23:56 I need to search for AI startups using my web searching tool. Okay, let's do
23:59 that first. Then I'll need to create a new spreadsheet with my Google Sheets
24:03 tool. And then for each company that I find, I'll need to add a row to the
24:06 spreadsheet. And then I'll need to write a summary of each business and put it in
24:09 a new column. And then finally, I'll use my email tools in order to send the link
24:12 to Liam. And that's all great, but then when you add on top of that powerful
24:16 reasoning models like OpenAI's 01 and 03 and even things like deepseat as the
24:20 brain of the agent that can plan, take actions, then reflect and then plan
24:24 again and so on. You have essentially created a truly intelligent AI that
24:28 solves problems and approaches them just like a human would. So, say for example,
24:32 the original plan was to use the web search tool to search for AI startups
24:35 raising money. Probably a terrible search term, but what if that doesn't
24:39 return any good results to the agent? Well, a human would go, damn, I need to
24:43 change my search term or maybe I need to try find a different method of finding
24:46 these companies on like LinkedIn or something. The latest in AI technology
24:49 like these reasoning models, it allows these agents to do this exact same kind
24:54 of reflection and replanning in order to achieve their objective. And this is
24:57 when you can really see why we call them digital workers because they can do
25:01 things like planning multiple steps. They will use different tools in a
25:05 sequence and even adjust their approach based on the results from those tools.
25:08 Now, I should mention that this technology isn't perfect yet, right? So,
25:12 these multi-step tasks are often unreliable and agents typically need
25:16 human supervision for more complex workflows. But things are moving
25:21 incredibly, incredibly fast. In fact, we're already seeing the next evolution,
25:24 which is multiple agents working together. Instead of just one agent
25:27 trying to do everything, you can have one main agent that you give orders to,
25:30 and then it can use all of the other agents underneath it as tools where it
25:34 can send specific instructions. Like underneath the main agent might be a
25:38 research agent, which is best at finding companies and has its own tools. Then
25:41 you have a writing agent that's really good at writing summaries. Then you have
25:44 an emailing agent, which has got all the emailing tools. And so each of these
25:47 agents can be specialized in their specific task with multiple tools. and
25:50 then they all work together to achieve a common goal. This is exactly what major
25:55 companies like HubSpot and Microsoft and Google are building towards. It's these
25:59 entire workforces of AI agents that can handle complex business processes
26:02 automatically. In the next chapter, I'll show you how to build AI agents like
26:05 this for yourself using no code tools. But first, we need to understand the
26:08 different ways that these agents can actually be used in the real
26:14 world. So, we understand how AI agents and tools work under the hood. Now,
26:18 great. If you don't, please go back and take some notes, right? You should by
26:21 now have a whole bunch of notes um from the stuff that we've covered already.
26:23 And this stuff that you're learning took me two years in order to learn and and
26:27 be able to apply effectively. So, you best believe it that it's going to take
26:30 you two to three watches before it all sinks in. So, if you're feeling a bit
26:33 lost and and overwhelmed, don't worry. That's how it feels with learning
26:36 anything new or how it should feel if you're learning something that's
26:38 actually pushing your boundaries and adding something to your to your
26:41 capabilities. Next, we need to look at the different ways that AI agents can be
26:45 used in the real world. There are two main categories of AI agents.
26:49 Conversational agents and automated agents. Conversational agents are ones
26:52 that humans interact with directly through chat on things like websites.
26:56 You've got maybe you're chatting to it on WhatsApp. You've got interacting with
27:00 it over the phone via phone call. You've got chatting to it via Instagram DMs or
27:04 custom apps and websites. For example, OpenAI's GPT platforms allows you to
27:08 create agents that you can chat with directly on your computer or on your
27:11 phone. or using platforms like my own Agent, you can connect these agents that
27:15 you build onto a WhatsApp number or onto Instagram. And I'll show you how to do
27:19 this in the tutorial chapter of this video. So, in all these cases, you or
27:23 someone else is there sending messages or instructions to the agent and
27:26 explaining what you want to do and kind of chatting back and forth with it,
27:29 whether it's on a website, WhatsApp, Instagram, or whatever. And within these
27:32 conversational agents, it's not just text based. It's like I said, there's AI
27:35 voice agents as well, which are an extremely exciting sector of the AI
27:39 space right now. And these systems use multimodal models that can take in audio
27:44 as input and then produce audio as an output. And so these agents can be
27:48 chatted to over the phone or via audio rather than via text. This AI voice
27:52 stuff is super cool. And in the tutorial section, I'm going to show you how to
27:55 take the exact same AI agent that we can chat to on a website and then connect it
27:58 to a phone number and talk to it on the phone. But then we get to what I call
28:02 automated agents. And so these are slightly different from the
28:04 conversational ones. The truth is that AI agents don't always need humans to
28:09 talk to them and use them directly. All they need is some kind of input or
28:13 instructions to trigger them and that tells them what to do. This means that
28:16 we can build these automated agents that instead of waiting for some kind of
28:20 human input, they are actually part of larger systems and processes and they're
28:24 triggered automatically by events like a new email received or a form submission
28:28 or they work on schedules like once a day and they essentially work in the
28:32 background without necessarily having human oversight or input. For example,
28:35 later in the video, we are going to be building an automated agent that is
28:39 triggered by a new form submission. When the form is submitted, some of that form
28:42 data is taken and sent to the agent, which then causes it to use the tools
28:46 that we've equipped it with and follows the instructions in the prompt that we
28:49 gave it in order to make decisions and take appropriate actions on our behalf
28:53 in a fully automated way. We are still sending the message to the agent, but
28:57 it's not a human needing to type it manually or speak it over the phone.
29:01 There's no human step. The input is being automated in some way. And this of
29:04 course opens up a huge number of use cases for AI agents in businesses
29:07 especially. And of course I'll be showing you how to build both types of
29:10 these conversational and automated agents in the tutorial section of this
29:14 video. But the last step of building your foundation of knowledge before we
29:17 move into that is to look at some real world examples of how businesses are
29:23 using these AI agents right now. So firstly we have the personal
29:27 assistant category. And this is what most people think of when they hear the
29:30 word agent. something that you can chat to that's going to update your calendar
29:33 and sort of send emails and even make phone calls for you. Um, now these are
29:38 all nice to have features, but honestly uh this space is likely going to be
29:41 dominated by the big tech giants. You've got OpenAI through Chatbt trying to do
29:45 this with Tasks, Google through their suite of apps and connecting them to
29:49 Gemini and Apple through Siri. These guys are going to eat up this entire
29:52 market of personal assistance and your own personal AI agent that helps you do
29:56 personal stuff. the real opportunity lies in business applications and how
30:00 people like you and I can build and sell AI agents to businesses which we're
30:03 going to be covering in depth in the final chapter of this video. So, we've
30:05 got the next chapter which is going to be on building the four tutorials and
30:09 the final chapter is all about how to sell and how to monetize your AI agent
30:13 skills that you've just learned. One of the core use cases for businesses right
30:16 now are what's called co-pilots. And these are AI agents made for specific
30:20 roles in a business. We're going to be building one of these later in the
30:23 video. And these specialized AI agents are essentially helping someone in a
30:27 specific role in a business to do their job more effectively. Take a customer
30:31 support co-pilot for example. It would have a knowledge base that allows reps
30:34 to get answers to customer queries instantly and deliver them over the
30:36 phone. So they've got the little co-pilot up on the side there. They're
30:40 on the phone. They get a question, they can search and for an answer in the
30:42 knowledge base, it gives them back and they can give it to them over the phone.
30:45 This same agent could also have a tool that allows them to look up the customer
30:49 information very quickly. Um, I could have another tool that it makes it very
30:52 easy to send a summary of the call into the database so that the next rep who
30:55 picks up the phone and talks with them knows exactly what was discussed
30:58 previously. It's like giving every support rep some kind of AI assistant
31:01 that makes them dramatically more effective. It also makes their customer
31:05 support a lot more consistent as to what the company wants people to be saying,
31:08 which is a a big problem with managing large customer support systems. And then
31:11 we have lead generation and appointment setting agents. These are probably the
31:15 most valuable type right now. And businesses are using these on their
31:18 websites, through WhatsApp, on Instagram, and even over the phone to
31:21 engage and have conversations with the interested people who are approaching
31:25 the business 24/7. They can offer instant answers about products and
31:28 services. And they're even smart enough to be able to capture emails and phone
31:32 numbers mid-con conversation for later follow-up by sales team. Some can even
31:36 book appointments on the spot and mid- conversation by using a tool to check
31:39 the calendar availability and then using another tool to create a new booking
31:43 once they've agreed on a time with the prospect. Another real world agent use
31:46 case and one of my favorites is a research agent. And so these can help
31:50 businesses by automatically researching leads that come in through their website
31:53 or elsewhere. And when someone fills out a form, the agent can spring into action
31:56 and start searching the web for information on the company, finding
31:59 their LinkedIn profile of the person they're going to get on a call with and
32:02 gathering any other valuable data that it can find. Then it can take all of
32:05 this information and generate a summary of who this person is and what this
32:09 company is also and decide whether they're a good fit for working with the
32:13 company and if so then they can send the sales team some kind of detailed brief
32:17 or suggested strategy on how to close this particular person on a call based
32:20 on the research. So it's basically like having a an automated team of
32:23 researchers who as soon as leads show interest in your business, they're out
32:26 there figuring out everything about them and determining one whether they're a
32:29 good fit for you and your products and services which is called qualification.
32:32 Then secondly, if they are qualified, giving the sales rep something that will
32:35 bring them up to speed on who this person or who this company is and how
32:39 they can try to close them. So, we have covered a lot so far.
32:43 So, before we dive into each of these agent builds that I'm going to walk you
32:47 through over my shoulder, please make sure that you've got your notes taken
32:50 out and the core concepts of this video so far understood properly. You should
32:54 be clear on things like what is the definition of an AI agent? What are the
32:59 five parts of an agent? How is building an AI agent like being a chef? And how
33:02 many ingredients do you have to play with? What are the two main parts of a
33:08 tool? And what do schemers do? So, pause the video now and try to answer these
33:11 questions. And if you aren't 100% confident, you need to go back and watch
33:15 it again. So, don't rush this or you're going to feel way out of depth when we
33:18 get into the tutorials that we're going to be covering next. But if you are,
33:21 congratulations. You are one step closer to AI literacy and becoming a much more
33:26 valuable uh participant in this global economy. So before we get into the
33:28 second chapter, there's just three very quick things from me. Firstly, if you
33:32 are a business owner who wants to fast track to becoming an AI leader within
33:36 your industry, at my agency, Morningside AI, we offer everything from AI
33:39 education and upskilling programs for executives and staff to AI strategy and
33:44 roadmap consulting and of course AI development services as well. So we
33:47 would love to help you get ahead. So feel free to get in touch via our
33:50 website in the description below. And secondly, at Morningside, we are hiring
33:53 for all sorts of roles right now. So whether you want to build AI systems for
33:56 some of the world's biggest companies that we have as clients or to help
34:00 produce videos like these that are seen by millions of people or create
34:04 educational material for thousands of businesses. Uh we have roles for all
34:06 sorts of things right now. So you can apply using the link in the description.
34:09 And please, even if you're just vaguely interested, I really recommend you just
34:12 check out the link and see what roles we're hiring for. Uh you never know
34:16 what's going to be on there. Um and it may be a very good way for you to use
34:18 your skills to fast track into the AI space by working under myself and my
34:22 team. And finally, if you have gotten any value so far in this video, please
34:25 head down below and leave a like on the video. It helps me reach more people.
34:28 Um, I put a lot of work into these videos and it also lets me know that you
34:31 enjoy this kind of content and that I should make more of it. And of course,
34:33 if you like this kind of content and want to see more of it, you can
34:36 subscribe so that YouTube will put my videos up for you whenever a new one is
34:39 released. So, there's also a little share button if you want to click that.
34:41 That'll let YouTube know this is good content and that you're sharing it to
34:44 other people. Not only will that help me, but you can share it to your friends
34:48 and family who may also or you may want to help them to brush up on these skills
34:51 or help them give a way to get on the front foot with AI. And that's what I
34:54 really make these videos for. So, thank you for sitting through that little bit
34:57 of housekeeping and self-promotion. Now, building. I have carefully assembled
35:05 this chapter on building to give you the most bang for your buck possible in
35:09 order to kick off your AI agent learning journey. We are going to be covering
35:12 four different use cases across four different AI agent building platforms.
35:16 These are all no code, so don't worry about that. And the chances of you
35:19 falling in love with at least one of these platforms is pretty much 100% as
35:23 you're going to rapidly start to connect the dots uh about how you can start to
35:27 use these kinds of agents and these platforms in your own life or in your
35:30 work or for your friends and family and those around you. So, here's a quick
35:33 rundown of the builds we're going to be getting into. The first build is going
35:36 to be a sales co-pilot built with relevance AI. And here we're going to be
35:40 building three custom research tools from scratch, including an advanced web
35:44 scraping tool, which is a a great skill that I want to teach you. And with
35:46 these, we are going to be creating a conversational agent to help the sales
35:51 reps at Big Boy Recruits, a hypothetical fantasy uh recruitment firm, in order
35:55 for them to be better prepared for sales course. So that's the purpose of the
35:58 sales co-pilot. The second build is going to be an automated lead
36:01 qualification agent. And this will be built on a platform called N8N. And this
36:05 time we will be helping Big Boy Recruits, our fantasy recruitment firm,
36:09 to automatically research and qualify new leads and then send an email
36:12 notification to the correct sales rep. And this is going to show you that
36:15 automated style of agent where it's built into a process rather than having
36:18 a human input necessarily. In build number three, we will be building a
36:22 website and phone-based lead generation customer support agent. This will be
36:25 built on voice flow and it's going to be able to do three things. Firstly, answer
36:28 questions from a knowledge base, generate instant quotes using a custom
36:33 tool we build and also do lead capture on interested prospects. We're then
36:36 going to slap this agent onto a website widget so that you can chat to it via a
36:40 website and via chat and text. And then we're going to take that exact same
36:43 agent and connect them to a phone line so that we can call our agent over the
36:45 phone and access all of the same functionality we just talked about. And
36:49 finally, for build number four, we'll be using my own AI agent platform, Agent,
36:53 to rapidly build a lead generation agent and connect it to a WhatsApp number that
36:56 we can chat to. The leads that we collect are going to be automatically
36:59 sent into an Air Table database for later review. And please don't skip
37:02 around these builds as they're all kind of connected in some way where we're
37:05 reusing parts from build one and build two, etc. But without further ado, let's
37:09 get into building some agents. All right, people. Enough of the theory. Uh,
37:14 now we get into the fun bit of actually building these agents out. So, I've done
37:18 a lot of work and my team has done a lot of work. So, thank you to the my team
37:20 members who have helped me put this together. Um, putting together four
37:25 different AI agent builds for you. And this is really going to walk you through
37:28 an A to Z all the different platforms that you really need to care about, all
37:32 the different kind of core use cases and functionality. There's a lot more of
37:35 course, but this is going to really give you the foundation that you need to
37:39 succeed in the space. And hopefully it'll be the thing that kind of sparks
37:43 your interest in it because I I want you guys to have fun with it. these big
37:46 tutorials for me. Honestly, when I put a lot of work into it, I build up the sort
37:49 of mental resistance to it because I know how much work there is going into
37:52 it and I have to make this big whole session where I'm all uptight about it.
37:55 But I'm just going to try and relax and enjoy this. And I really want you all to
37:58 do the same. So, set a bit of time aside. You can either pause this video,
38:02 put on your watch later, but I really want you to take your time with this.
38:06 I'm going to be doing this more. So, when you do tutorials like this, there's
38:09 a few different ways you can do it. I can either do all the building and then
38:12 give you the templates and kind of just spoon feed it to you. And that's more so
38:15 what you do for someone if you're trying to like really fast track them and they
38:18 don't want to learn all the skills, but um I I know what I'm trying to build
38:21 here for you guys. And I'm going to give you a sort of stream of consciousness.
38:25 You just get to see me kind of jamming out and building these things. And I'll
38:28 be explaining my thought process and the concepts etc along the way to reinforce
38:33 what we've learned before. So I'm just going to dive into it with our first
38:40 And so what I've done is put together a big Figma board here which is going to
38:42 be breaking down all these different builds. So under here there's some
38:45 goodies you see. Oh, there's some goodies under each of these that I've
38:48 put together. Um and we're going to go through them one by one. Starting off
38:51 with agent one over here. I mean there's a lot of stuff here um that you guys are
38:54 going to get. So you'll get the whole Figma and it includes all the templates.
38:57 So if you do want to just kind of watch through this, pick it up. You can either
39:00 do it and follow it step by step with me and see how I build it and really build
39:03 those flexible skills that you're going to need to succeed in the space or you
39:07 can just watch it and be like, "Okay, I kind of get what he's doing and then
39:10 take all the templates from me at the end." That's I mean, completely up to
39:13 you. Depends if you want to be a really really nerdy builder about it and get
39:17 into the weed like like I like to do. Um or you just want to be like, "Hey, I
39:19 want to do this my business. I want to roughly understand how these things work
39:23 and what platforms." So, use this resource as you will. But we're going to
39:26 jump into agent build number one here, which is our sales co-pilot built with
39:30 relevance AI. So, running through this quickly, we have the purpose of this.
39:33 This is basically going to look a bit like this. It's going to be a co-pilot
39:37 and co-pilots work in that you have a uh it's basically a specific AI agent that
39:42 you build for a specific staff or staff member or role. So, say this case, it's
39:46 going to be a sales co-pilot. It'll be the thing that the sales rep uses to uh
39:50 in their day-to-day as they're working on their jobs. You can add tools like in
39:54 this case, you see we're going to have three different tools here for our
39:56 agent. One's going to be a company researcher tool. So this is when the
40:00 sales rep would be like, hey, I have a call coming up soon. Um, let's put in
40:03 this I need to research this company cuz this is who I'm going to be on a call
40:06 with. So they'll put in the company URL. This tool that we're going to create is
40:08 going to go and research that company. It's going to bring back and give a
40:11 summary. And then it's like, okay, well this is the LinkedIn URL of the person
40:14 that we're going to be got on a call with shortly. It's going to pass in the
40:17 LinkedIn URL. It's going to take that URL. It's going to pull all the
40:20 information and write a summary about the person. So now we have the company
40:23 summary and we have the person summary. And the final step here is going to be
40:26 what I'm calling a pre-all report generator. And that's going to take both
40:29 that company and prospect research that we've done. It's going to combine them
40:34 together and be for this specific company. As you're going to see that
40:37 this hypothetical company we're building this sales co-pilot for, it's going to
40:41 generate a basically a pre-core report or a strategy uh a strategy prep for the
40:44 sales rep so that they go onto those calls much more prepared and also sort
40:49 of a personalized guide on how to try to close this person. So, um, all of these
40:53 templates are going to be here. Each of these are templates for the tools. And
40:57 this is for the agent as well. Um, but here's some more information. You guys
40:59 can pick through this as you wish. But, um, this is the kind of end result and
41:02 we're going to be able to chat to it. And this would be something you could
41:04 build for a client. You could build it for your own business or you could just
41:07 tinker around. You could build co-pilots like this on relevance for yourself. So,
41:10 that's why I want to start with relevance because it also is a platform
41:13 that we can build these tools on. So, it's a really, really good one to start
41:16 with and let's get into it. So, the first step of course is to go to
41:19 relevance AI. So, I'll put a little link up here. You guys will be able to get
41:22 this Figma. It'll be on the school. Um, all of the information and all the
41:24 resources for this are going to be like this Figma is going to be linked to the
41:27 school. My free school community. If you haven't already joined, biggest AI
41:31 community on school, biggest AI business community probably in the whole world.
41:35 Um, so we can jump across that first link in the description. You'll be able
41:37 to find this in the YouTube resources section. Um, pretty straightforward. Of
41:41 course, when you click on this, it's going to ask you to log into relevance.
41:44 So, if you haven't already, you can make an account. Um, it's fairly low cost.
41:48 They have a free plan, then a team's plan, I believe. Um, so it's not too
41:50 much, but it is a really, really valuable tool as you're going to see.
41:54 So, you can sign in here. I'm going to jump in with my Google account. There
41:58 may be a bit of setup for your account, but I'm sure you guys are smart enough
42:01 to figure out how to set up an account. I'm sure relevance also makes it easy
42:04 enough. So, then we get taken to this dashboard, but we see on this left hand
42:07 side, we have tools. So these are the tools that I've talked about um where
42:10 it's some kind of functionality that we can create and we can build it all on
42:13 relevance no code and there's even sort of extensibility or you can add more
42:18 functionality in to relevance by adding some low code components or even custom
42:21 code. So relevance is a really great base for building not only tools but
42:24 then the agents that you can connect that into and we're going to use the
42:27 same relevance tools that we make now in multiple of the different agents that
42:29 I'm going to make for the rest of this video. So first things first if we go
42:32 back to the Figma here we see we need to make three different tools. So tool one
42:35 is company researcher. It's going to take in a company URL. It's going to
42:37 search the web and it's going to return a summary. So, that's the functionality
42:42 we need. Let's go and create a new tool. Going to call this um
42:48 research company. We can give it a cool little I'm going to zoom this up for you
42:51 guys. Hopefully, that's the right size. Um uh search. Have fun with this stuff,
42:57 guys. Like, if if it's putting stupid emojis on things to to enjoy it, then
43:01 then that's what you need to do. Like, uh if you make it a chore, it's going to
43:04 feel like a chore, right? Um, now we get to descriptions. Um, this is something
43:07 we're going to see recurring, but basically as as you know, as we learned
43:11 in the concept section, we need to have natural language descriptions of our
43:15 tools and of our APIs so that the agents can read those descriptions and
43:19 understand uh what the agent or what the tool does and what those different parts
43:22 does. So, you'll see this recurring throughout this. But first things first,
43:26 what does this research company tool do? Um, takes and oh, I got caps lock on.
43:32 takes um and there's also some tutorials here if you want to go deeper. Relevance
43:36 has some great documentation as well. Um but takes in a company
43:45 penny URL and scrapes the website then returns a sum and AI generated summary. So then to
43:57 build a tool we need to have some kind of input. You don't always need an
43:59 input. It can actually just be triggering, but generally you're going
44:02 to have some kind of input that the agent needs to pass into the tool. In
44:05 this case, to research company, it's going to need to take in some text,
44:09 which is going to be a URL. Um, we're going to say comp company
44:15 URL. And then again, here we have another description. You see, describe
44:18 how to fill this input. This is again going to help our agent within relevance
44:22 AI and elsewhere as you'll see in tutorial number four. Why this is so
44:25 important to add in the descriptions, right? So this is a URL for a company to be
44:39 researched must be in the format https colon slash um dot dot dot dot dot. So we need to
44:48 have the https for this to work. So that's going to be our input. Now if
44:50 this seems a bit confusing just stick with me. It will make sense in a second.
44:53 this stuff. If there's anything that I've learned from picking up so many
44:56 different tools, like when I first got into Facebook ads, when I've got into
44:59 building these kinds of agents, it's you feel completely overwhelmed, but that's
45:02 all just part of the process. And what feels difficult now is not going to feel
45:05 difficult forever. So, just please stick with me. Um, and it's a really, really
45:09 great feeling once you go and be like, this was hard a few weeks ago and now
45:12 it's really easy. So, we've got our first input. That's what the research
45:15 company tool is going to do. And then we need to define our steps. So, the next
45:19 step, we can go add. I'm going to hide this so we get a bit more space. Add
45:22 step. Now, the cool thing about relevance is that it comes with a lot of
45:25 great functionality out of the box. Here's one, extract website content, we
45:30 have LLMs, we have Google searching, we have all sorts of AI generations,
45:35 replicate um knowledge bases as well. There's so much cool stuff on here and
45:38 this is why I really rate relevance as one of the best platforms. If you were
45:42 to go all in and want to upskill, you can build so much on this. Um, so I'm a
45:45 big fan. I love the the relevance team and what they're doing. So the initial
45:48 plan for this build I was going to use the extract website content which is
45:52 fairly straightforward. We can say oh one other thing um the company URL we
45:56 are going to be using this company URL throughout our tool here. So we can
46:00 change this text to say something more descriptive. So company
46:06 u URL. You guys are going to got going to think like think and write like
46:09 coders now cuz you need to uh use some kind of syntax and use some kind of
46:13 variable naming convention. This is a a standard one or you can do things like
46:18 company URL camel case but I prefer this format as I'm I'm mainly a Python kind
46:21 of guy myself. Now that we have that named we can use it in these kinds of
46:25 fields. So here you can see pops up use inputs. So basically when this tool is
46:29 run it's going to take the inputs or the information we put in the inputs and
46:32 it's going to pass it to different steps and use them as we describe within the
46:35 within the builder here. So let's run morningings.ai. Um, and then we can
46:49 click uh run here. So, that's going to go to my website and scrape the
46:54 information off of it. There we go. So, it's got all of this, but you see it's
46:59 just pulling back the first page. Um, and this is why I actually I shuffled
47:02 this around. And I want to show you guys how to do something a bit more advanced.
47:04 I know this is supposed to be a beginner tutorial, but this is not really that
47:08 useful and I it's a very easy thing for me to just bump this up to a little bit
47:12 more valuable. Um, while relevance tool here is great, we can do better. So,
47:16 we're actually going to delete this. Um, you can use that step for all sorts of
47:20 other things, but I really like what's called fire crawl firecroll web scraper. This is a cool
47:29 app. Um, firecrawl.dev. Shout out to the guys at firecraw. Basically, if we then
47:33 go HTT Oh, I should just see if they can doti and now do a free scrape for us
47:42 here. So, this is just going to do the single URL just like we got in
47:46 relevance. But the difference here is if we then go to crawl, if you hover over
47:51 this, it's going to crawl a URL and all of its accessible subpages outputting
47:55 the content from each page. So, instead of just taking that front page, it's
47:57 actually going to crawl through multiple things. So this is really the first cool
48:01 thing or or first skill that I want to put in your tool belt is that you have
48:04 things like fire call that you can use their relevance. They have things like
48:06 map which is just going to output all the URLs that it finds. Then there's
48:10 other things here where you can use AI to extract data. I'm not going to go
48:12 into that. But what we want is this crawling functionality from firecraw. So
48:17 I'll put a link in on the school post that comes with this video. It again
48:19 first link in the description to go to the school and if you go to the YouTube
48:23 resources section um there will be a uh a whole post on this and all the
48:26 resources will be in there. It'll also be in my free course on school as well.
48:29 So you can find it in the classroom section. So you want to sign up to FCL
48:34 so you can get API key. It's very easy. We can just go through with Google.
48:37 Again, this is not sponsored and there is zero sponsoring going on through any
48:40 of these tools. I guess I'm kind of sponsoring my own tool because I'm
48:43 putting it at the end. But I'm not getting paid a dime for any of this. I'm
48:45 really just trying to put you guys on what I like to use, what's made me
48:49 money, what's made me a more valuable AI automation expert or developer for my
48:53 companies and for the companies you work with. continue and then you get to the
48:57 dashboard here. It might look a bit scary, but what you can do is go to the
49:00 API keys. So, you can click on create an API key here, YouTube. I'm having issues with mine. It
49:06 should be fine for you. I've already got an API key, but once you get the API
49:09 key, what you can do is take it and come back here to relevance. And you can go into the side
49:16 panel here. Where is it? Settings. And then we have our API keys. So, we're
49:19 going to need to add more into this later. So, keep an eye on this. This is
49:22 something you need to be familiar with. Um, and you can scroll down to firecrawl
49:25 and you just pop it into this firecall API key section here and you're good to
49:29 go. And you can come back. Oh no, we don't want to duplicate
49:35 that. Now we've got our firewall set up. We need to make a couple things. We need
49:39 to do a couple tweaks here. We have, of course, the if you want to get those
49:43 variables up, you can go bracket bracket um or curly bracket curly bracket, which
49:46 is shift um to get those. I don't know why it's not popping up. There we go.
49:51 Company URL. And if we hover over this, we can see it says scrape the provided
49:54 URL only. Uncheck if we want to crawl instead. So if we want to get that crawl
49:58 functionality that we just saw that we think we want to get all the data, we
50:02 can uh uncheck it. And then we want to extract the main content. So you might have to just trust
50:07 me on that one that we don't want to have all of the other rubbish. We just
50:11 want the body of the website. Um a number of pages. We don't want this to
50:13 take too long. You can expand this much more. Um but I'm just going to go for
50:18 say five for now just to keep it uh keep it quick. And then now what we can do is run this
50:24 again. We've still got my URL up here. We can go run step. Give it a second. You will at some
50:31 point have to pay firecrol. Of course, it's not a free service, but they do
50:33 have a free plan, so you should have no issues with getting that. So here you
50:35 can see we're getting a lot more data back from this web scrape than we were
50:39 with just the relevance version, which is great. So the next step is we have
50:42 this data. We want to generate some kind of research summary um so that we can
50:46 send that to our sales script once they have requested it. So now it gets into
50:51 the fun part of writing LLM prompts. For this one, sometimes you need to really
50:55 go and and make a big effort, which we are going to do later as you'll see. But
50:57 in this case, we just want a quick summary. I'm going to throw one in here
51:00 that I made earlier. All of this will be available on the Figma or it'll be given
51:03 somewhere in the resources, right? But um if we just need something quick and
51:06 dirty here, it's not really a massive part of the project, so it's okay to
51:09 just whack one in there. So bang, I've got it in there. Can you please take
51:12 this website content and summarize it into a 300word natural language summary,
51:15 which clearly outlines rad where they're based, their values, etc. anything that
51:19 would be helpful to know for a sales rep who will soon be on a call with them. Uh
51:22 break it into key areas like overview, products and services team etc. And I've
51:25 got a couple things here to make sure that it doesn't mess up which I it was
51:29 doing for me a bit in testing. So we do want to put in if we go curly bracket
51:32 curly bracket we want to put in the fire call data here which is going to be uh
51:36 all the website data that came back from our scrape. We want to then insert it
51:39 into this prompt here. So I hope you're starting to the things in your your cogs
51:42 and your brain are starting to click into gear here. And then we get to
51:45 select the model for this. It's pretty basic task. So, I want something quick
51:48 and cheap. Um, for many of you, it's going to be easiest to work with the
51:51 Open AI APIs because you've probably already played around with your API key
51:54 before, which I'll show you how to do in a second. But, let's go with GPT40 Mini,
51:58 but Relevance, of course, does have support for all of uh the other models.
52:02 But my tendency for most tasks now is to actually go for some of the Google
52:05 models that have come out. Again, you guys might be watching this in a year or
52:08 whatever. It might be quite different, but at the moment, Google is really
52:11 leading the way with making the cheapest models possible, and they're actually
52:14 really good as well. the the price decrease on using things like uh Google
52:17 Gemini Flash Light and Google Gemini Flash 2.0 and stuff like that. It's
52:21 ridiculously cheap and it's also a really good model. So, it's a bit more
52:24 difficult to get APIs on the Google side. So, I'm just going to stick with
52:27 OpenAI for this tutorial. Um, so let's go GPT4 mini. And then we need to of
52:31 course set up our API key. Um, if you scroll down, where is the OpenAI? So, to
52:39 platform.openai.com.playground/playgroundplayground, sorry. Um, this will again be linked in
52:43 the resources or you can just type up platform.openai.com. I'm sure you'll
52:46 find it. If you haven't already, create an account um, and sign in. And then you
52:50 need to go to your dashboard here. You go on the left side, you go to API keys,
52:54 and you click create a new secret key. And then you're going to be able to copy
52:57 that key and bring it back into uh, relevance. Paste it in here. And then
53:00 you're good to go and start using the OpenAI suite of models. It's pretty
53:04 easy, right? So now we have our information back from the scrape. We
53:09 have our prompt in here. And then we can go run step and see what this company
53:12 research tool is going to output for us. Boom. Morningside AI is a leading
53:15 artificial intelligence development company dedicated to empowering
53:18 businesses autonomous AI agent development, enterprise consulting,
53:23 chatbot development team. Uh they got keep me out of my own team page.
53:27 Damn. But uh yeah, so there you go. That's the that's the company research
53:31 tool. That's step one. Um I hope you guys can kind of see how that works. Now
53:33 a cool thing about relevance is there is this build section which we've just gone
53:36 through. But you can also go to use and this can be really helpful when sharing
53:39 these kinds of tools. So not this is not really only just an AI agent tutorial.
53:42 I'm also teaching you how to build tools because you can build very valuable
53:45 tools and something like relevance. And then you can go share. You can go
53:49 publicly available. Oh, and I can click this and then I can give to my employees. I can
53:56 give to the companies that I'm working with, the clients that I've sold these
53:59 kind of services on. And then they get a nice and handy tool like this. I mean, I
54:02 use these throughout my organization for like description generation, a lot of
54:06 content repurposing. There's tons and tons of different use cases for building
54:08 these kinds of tools. And then you can take this URL and you can share it
54:11 around to whoever you want. So, this is a uh a great way to use relevance. Let's
54:15 go back to our tool here. But for now, we need to keep moving along so we can
54:18 get this this first agent done. So, that's the first tool. I'm going to run
54:20 through it a bit quicker now that we know how these kind of tool buildings
54:24 work. I'm going to create a new tool here. I'm going to call it this time if
54:28 we go back to our Figma. And I recommend when you guys are building your agents,
54:30 you're building systems and planning them out. This kind of laying out in a
54:34 Figma, if you're not familiar, this is Figma. It's like a design software, but
54:36 you can also use it for kind of whiteboarding. It's called a Fig Jam
54:39 board. It's one of the types of uh boards you can do. And I use this all
54:43 the time with my team as well. It's like if you can't take what I'm telling you,
54:46 lay it out on a board so that I can review it and give you notes and and
54:50 then we all agree this is the build. Um there's often a lot of uh communication
54:53 issues with explaining functionality. So this AI agent layout, it's really
54:57 helpful. You can if it's maybe a workflow automation, you can do box box
55:01 arrow arrow arrow all of this laying out how it's going to be built. Here I've
55:04 done it for you in a very basic format. So we've done this first one. Maybe I
55:07 just make this green. Um second one is going to be prospect researcher. So this
55:11 takes in a LinkedIn URL, searches the web, and returns a summary. And you're
55:14 going to see how I tie this all into an agent shortly. So we're going to go research
55:22 prospect. Sil takes in a linked in URL scrapes the profile and then generates
55:28 an of the prospect input. We're going to need a link and URL the link linked and
55:35 URL of the prospect. this. Now, we're going to add a step and
55:46 relevance has got us here linked um get a LinkedIn profile or company
55:53 post. So, this is cool cuz then we can pop in our LinkedIn URL
56:03 them. So, we're going to get the user mine my LinkedIn profile, if you guys
56:12 want to connect with me on LinkedIn, more than welcome to do so. I'll put in
56:17 the description below. Um, we can do a little run step here. So, if we go back
56:22 over to data here, that's great. We've got my about section. So, this goes very
56:26 long way across because it's in uh it's in JSON here. It's got my company. It's
56:33 got my company domain, where I'm from, years, company, founded, tons and tons
56:36 of great information that you guys can use and we are going to use shortly. So,
56:39 this is really cool. I would probably add one more step to this if I was
56:42 taking this um and building it for for my own team. I would add in another
56:47 LinkedIn scrape here um where we just do the same thing, but we also get the
56:50 posts because posts can give you a bit more up to date um information on what
56:53 they've been doing recently. that you can guys can add that and you just go
56:57 add a step LinkedIn and you do the same thing as we've done here but you change
57:02 this to LinkedIn post get user post so that may be a cool thing for you guys to
57:05 add the uh functionality on at the end is a bit of a challenge you can pause
57:07 this video and do that and then we need a llm step to take this again I'm going
57:12 to grab a pre-written prompt that I did just to save some time going to drop
57:17 this in here says fairly similar stuff um LinkedIn data I'm going to put all
57:23 the data in there. And then we're going to use a GPT4 mini again. And we can give that a
57:30 run because we've already got this data queued up here. And there we go. We've
57:34 got a nice summary. Um, if I change this to nice and formatted. So there's a
57:38 little button down there between raw or formatted. And you can copy the stuff
57:41 out of here, of course. So Liam Mley, my followers. Damn, I got a lot more than I
57:44 thought. So there we have the summary, my name, where I'm based, uh,
57:48 information, my career experience. Super handy stuff. And this is going to be
57:51 super helpful in the next step when we generate that pre-core report. So very
57:55 quickly, we've created one more tool. I'm going to save this. So now we've
57:59 got, if we go back to our Figma, we have two of these done. Now the final one is
58:03 going to take in the company and prospect research and generate a
58:06 pre-call report. So this one's going to be a little bit different. If I go
58:11 create a new tool, preall report tool. Okay, free call takes in company
58:21 and prospect summary and generates a free call report for sales
58:26 direct. And now for the inputs for this, we need long text, not just normal text
58:29 because we're going to be taking in that big company and uh and prospect
58:34 research. So we go prospect summary summary of the prospect based on length
58:43 that and a prospect in this case is someone who's a potential customer. Just
58:47 to clarify that if you're new to business and don't really get these
58:58 summary, right? We have our prospect summary and company summary in there. I
59:01 hope you're following along. Next step is just an LLM step and we want to
59:04 combine these two together. Again, I've got a little handy prompt for this to
59:07 save us time. Now, in this case, you will see that the prompt is a bit
59:11 bigger, right? So this is um for more important parts when you are creating
59:14 tools whether it's it's for agents or just generally when you're using prompt
59:17 engineering and LLMs to create value. In this case we are creating value because
59:21 in this case we're taking in this prospect summary and this company
59:24 summary. We're also giving it the the context of this fantasy or or
59:28 hypothetical business that we are selling this agent to as like a co-pilot
59:32 system. We've got Big Boy Recruits which is Dallas based recruitment firm
59:35 specializing in software industry talent acquisition for SMBs. You're going to
59:38 see this kind of recur across the different projects we do, but basically
59:41 we're helping these big boy recruits to automate their business with AI. So,
59:45 this takes in some context on that business and it's going to generate a a
59:48 report that's going to help the sales rep say, "Okay, this is the company.
59:51 This is what we sell. This is what we specialize in. This is the company that
59:54 we're trying to sell to. This is who we're going to be talking to. What's
59:57 some how can I personalize this call or what's the strategy I can go into this
60:00 with? What are some angles that I can attack this call from?" And so to do
60:04 this, I have a prompt writing tool that I use quite regularly and my team uses
60:09 it as well. Um, perfect prompt. So, this tool does a lot of leg
60:13 work for myself and my team all the time. I'm going to give it to you guys
60:15 to use for free. You'll be able to clone it into your relevance account.
60:18 Basically, what I'll usually do is I'll put on uh the dictation thing. You've
60:22 got it on one of these keyboards. You've got this little thing. basically
60:26 whatever on your computer allows you to speak into the computer and it takes in
60:30 your voice and transcribes it into into text on the screen. I'll press that and
60:33 then I'll explain as you can see here what is this prompt doing and why and
60:36 I'll go this prompt needs to do this this this is going to take in this
60:39 information it's going to do this the reason we're doing this is this this and
60:43 I'll do like a big big body of text in there and then the next one if I have
60:46 them I'll give some good examples of input and output pairs of how I want it
60:49 to take in data and how I want it to spit it out. If you give it both of
60:52 these and you hit run, it's going to print you out using the researchbacked
60:55 prompting techniques that we use at Morningside. It's all crammed in here.
60:59 There's a video that I recommend all of you watch. It's going to be in the free
61:01 course anyway on school. So, when you get in there and watch my prompt
61:04 engineering guide, um, this is basically the entire information of that prompt
61:08 engineering guide smashed into this LLM step here. So, when you pass this
61:10 information in, it applies all of that and it gives you out a prompt that is
61:14 fully researched back and performs very, very well right out of the box. So,
61:16 that's a little bit of extra value I wanted to throw in there for you guys.
61:19 this is going to be available on the school um with the rest of the resources
61:22 as well. So basically I put in the information here about what this
61:26 particular uh task was. I said it's going to take in the prospect
61:28 information. It's going to take in the prospect summary and the company
61:31 information and it spat out this basically first go and I just had to
61:34 insert these variables. So this prompt and everything else will be on the on
61:37 the score resources as well. So in this case because we are doing a bit of
61:40 strategy and sort of high level thinking rather than just summarizing and we may
61:43 want to change the model here to something a bit smarter. We could go to
61:47 03 mini which is one of the later ones. Um, again, when you're watching this, it
61:50 might be 06 or 010 or whatever the hell they come up with next, but there's
61:53 probably going to be some much better models. So, just use a smart one because
61:57 it's really strategizing on how big boy recruits can position themselves for
62:00 this call. So, enough of me yapping about that. Let me grab some inputs for
62:06 [Music] it. Shout out muscle. All right. So, I've got this information here about
62:12 myself, my LinkedIn profile, and my company. And so again, remember that
62:15 this is for Big Boy Recruits, a Dallas recruitment firm specialized in software
62:18 industry talent. So it's going to look at my company, Morningside AI. It's
62:21 going to look at me and my background and my profile on LinkedIn. Then it's
62:25 going to spin, as we see here, um, review this, analyze this, map big boys
62:29 unique value proposition, ra, and it's going to try and create a report that's
62:33 going to allow the sales rep to sell me or close me better on there or find some
62:37 angles to sell to at least. So if we go run tool, and now you see that I'm using
62:40 a lot of just basic web scraping and LLM steps. I just want to show you guys the
62:43 basics. The thing is tools can get very very advanced when you have like CRM you
62:47 want to integrate into, but relevance allows you to do all of that. It's just
62:50 within the scope of tutorial, it can be pretty difficult to be pulling
62:52 information from all over the here cuz I have to set up a database, show you guys
62:56 how to do it, too. So, this this keeps it quite confined, but it still gives
62:59 you a good taste of it. So, if we look at this view, all key business
63:03 challenges and opportunity, Morningside AI, this Liam's profile gives me a bit
63:07 of a rundown of this mapping big boy recruits unique value proposition. Maybe
63:10 it's going to be better if I change this to the format. There we go. Talks about
63:15 mapping big boys recruits, strategic talking points. I've been following your
63:20 journey of digital marketing, AI, ra um dive into opportunity. I work at big
63:23 boys. This assist, and it's even gone and done a section on anticipated
63:26 objections. So, the idea is that the sales rep is going to have a skimmer of
63:29 this before the call, which ties into the value that I've listed here, which
63:32 I'm going to do for all of these builds, by the way, which comes down to
63:35 ultimately a better prepared sales rep should close more deals, right? if they
63:38 know more about the prospect and the company and you have an angle to try
63:42 sell through or suggestions at least. It should increase the conversion rate of
63:45 the sales team. So, we've built this tool. We can change this to green now.
63:48 And the final step is going to be heading over to our agent builder within
63:52 relevance. I'm going to save this. If you pop over in the left panel here, you
63:56 can go into our agents. And what we want to do is create a new
64:03 agent. We're going to call it our sales co-pilot. Um, big boy big boy sales co-pilot. Sorry, I
64:12 got, like I said, I got to have fun with the stuff where I go kind of insane. Um,
64:18 this this agent is our sales co-pilot that helps reps to be better prepared for sales
64:30 [Music] calls. Triggers, we don't need to do any of that. We go to core instructions. I'm
64:35 going to again paste in some of the stuff that I've prepped earlier. So, if
64:37 I paste this in here, you'll see that it's structured fairly similarly to the
64:41 prompt that we just did before for the uh pre-core report generator. And this
64:45 is again using another tool that I've created um for AI agent prompting. Um
64:49 so, it's fairly similar stuff that I I include in that other prompting tool,
64:53 but the agents is slightly different um to just regular LLM steps within
64:56 different tools and workflow automations. So, I will include this as
65:00 well. It's my AI agent perfect prompt generator and it's fairly
65:02 straightforward to use. I'll include that in there as well. But basically,
65:05 you put in all the information about what the agent is, why it's doing it,
65:07 the different tools that you're connecting to it, and then it prints out
65:10 this for you. So, I'll just run through it. We've got a role here telling it who
65:13 it is, um, and kind of hyping it up and saying how good it is. Explains the
65:17 task, um, talking about how it's helping to conduct detailed research on
65:20 companies and prospects. Um, some specifics. Uh, don't need to worry about
65:23 those too much. Just reiterating the task. And now here we can enter in the
65:27 references to tools. So if we go slash tool and see in order to get the agent
65:31 to function as well as possible, we need to tell it what tools it has available.
65:35 This is really key across all the agents you build, especially if they're more
65:38 conversational. You need to explain to them what tools they have and how and
65:41 when they should use them. So if I go, company. Um, yep, that's right. Purpose,
65:50 input, company URL, use when needing to gather company information. This one of
66:01 prospect. And then this last one is our [Music] pre. There we go. So that's all whacked
66:05 in there. Don't need to worry about that too much. But again, this will be
66:08 included um this prompt and everything if you want to follow along and also if
66:11 you just want to clone the whole agent um and use it in your own business or
66:14 sell it or whatever you want to do. We also get to select the model here. Um
66:18 I'm just going to keep it as GPT4 mini. I like some pretty fast responses here
66:21 because agents as someone's using it, it can feel really irritating if it's not
66:25 responding quickly. So, we've got all that built out. That's the core
66:29 instructions in the prompt of the agent. Um, we can go down to the uh tools
66:33 section. It's got all the tools connected in here because we mentioned
66:36 them in the prompt. And we can just go through and do some quick settings on
66:39 here. Um, I don't want to have to do an approval for it. Some tools you can say,
66:43 look, they've got to give it a thumbs up before it can actually uh trigger it.
66:46 Um, prompt for how to use. Just some quick descriptions we can pop in here.
66:49 I'm not sure why relevance doesn't carry that over from the tool. I guess they're
66:52 asking us to do it again for some reason. Um when you need to research a
66:55 uh prospect linked and URL, we're going to say this is auto run as well. Um and
67:05 then preall change this to auto run as well. Use this when you need to generate a pre
67:15 call report from the company and research. All righty. Um, and there's
67:21 all sorts of other cool stuff. Relevance, as you can see, is like
67:25 abilities, sub aents, metadata, extra stuff that you can build onto. Um, but I
67:27 just want to get you guys started with the core of this. Um, all right. So, we
67:36 that. And boom, we have our agent here ready to go. So this is where you can
67:38 test your agents and use them if you want to. But in this case, I'm just
67:42 going to give it a quick rundown and say see if the functionality is working as
67:47 we as we planned. Um, hi, I am getting on a call with Liam Otley from morning
67:55 side AI. Here uh has lenol report please to prep for the call. Sending your task to big boy sales
68:24 copilot. And then we get to see all the debug and how it's actually walking
68:28 through these different steps. Oh, let's does. Yep. Okay, that's great. It's
68:36 using the research company as we wanted it to. Should add in one more step there to
68:42 research the prospect as well. There we go. Using the second
68:46 tool. It is pretty satisfying when this stuff works. And this is just a really
68:49 basic one, guys. I don't want you to think this is like, oh, well, that's
68:52 pretty underwhelming, Liam. I'm trying to teach you the basics so that you can
68:56 actually build on top of this. So if you get the bug, if you get the like you get
69:00 a travel bug, if you get the agent bug and you see the stuff and you're really
69:03 interested. Oh, look, there it is. Now it's filling out the prospect summary as
69:06 the inputs. Surely we don't have to watch it do that. It's going to take a
69:11 while. [Music] Um, I really write that out word for word like that. But when you start to
69:18 see this magic and you add in other cool tools and functionality, you test it on
69:21 yourself. You can build like things for maybe you want to do content, you make
69:25 yourself a little content co-pilot, etc. There we go. To use all the tools
69:29 and it should be spitting back and bam, there we go. So, I hope that was worth
69:33 the wait. Let's go through it now. Here's your comprehensive precore report
69:35 for your upcoming conversation with Liam Mley from Morningside AI. Pre-core
69:39 report. Talent acquisition under pressure. Morningside AI operates in a
69:42 highly competitive AI and tech market as they scale. Finding specialized talent,
69:46 engineers, data scientists, AI with a proven record can be challenging. That's
69:50 scarily accurate because that is literally one of the biggest constraints
69:53 that we have had to scaling Morningside long-term is that it's just really
69:57 really hard even with my channel. It's so hard to find the right people and get
70:00 them to commit as as developers as well. So if you want to build a very big
70:03 general AI development firm, an AI automation agency, you need the best
70:06 talent and you need to get a lot of it in um so that you can scale up. So
70:10 that's bloody spot on. Obviously this thing knows that's a good angle to sell
70:15 through. Um but yeah, prospect analysis. Liam's a dynamic entrepreneur and
70:17 thought leader with a robust background in e-commerce, digital marketing and AI.
70:22 His journey reflects a passion for innovation and commitment to continuous
70:26 learning. Again, that's pretty bloody spot on um hands-on experience. So there
70:31 you go. That is the big boy sales co-pilot for big boy recruits. The cool
70:34 thing you can do now once you have built this um is you can go share um there's a
70:40 chat UI which I'm going to turn on now. There are chat widgets so you can put
70:43 them on websites and stuff like that. What I want to do is just pull this up
70:46 because this is what you'd be giving to your client likely or if you guys are
70:49 going to start selling these to businesses which again we're touching on
70:52 selling in the last section of this video. So how do you turn these into
70:55 into a business and start making money from it and selling these as a as a
70:58 service and building these businesses which is really where the money's money
71:01 is made. So there you go. This URL you can obviously send to your client. If
71:03 you're building it for your own team, you can send this to your team and say,
71:06 "Hey, pin this because you're going to be able to use it. Add more
71:09 functionality into it, etc. That is how All righty, that is build number one out
71:20 of the way and we are jumping into AI agent build number two, which if you're
71:23 listening closely at the start of the section, we are talking about an
71:28 NATbased inbound lead qualification agent that's going to be doing some
71:31 pretty cool stuff for us, which is a really important function within a
71:34 business um around lead qualification. Um, so this is a really cool one. Again,
71:37 in this case, this is what we call an automated agent, not a conversational
71:40 one. What we just built is a conversational agent. humans are
71:43 directly talking to it and chatting back and forth and using it and we are
71:46 operating it ourselves. In this case, as you can see on this little flow uh flow
71:49 diagram here, this is a screenshot from the final product. We are actually
71:53 baking this into a workflow that's going to be triggered on a form submission.
71:56 We're going to do some research and then we're going to use the handy AI agent
72:00 and tools agent within NA10 to trigger another workflow and then send some
72:04 emails off. So, this capability of using AI agents in workflow automation really
72:08 expands the possibilities of what you can build. And the software NAN that I'm
72:11 going to teach you how to use is really at the the cutting edge and leading the
72:14 charge when it comes to these automated uh AI agent workflows. And just quickly,
72:17 it's good that we walk through the purpose and the value behind this
72:19 automation so that we know why we're doing it. Right? So this inbound lead
72:23 qualification use case is based on the fact that companies who market
72:25 themselves well soon have far too many people reaching out to them. Many of
72:28 which are not a good fit or what you'd call qualified for what they sell. Eg
72:32 they're too small or they're not the right industry. Like you you have a
72:36 business and they say we only help XYZ kinds of businesses. And if leads come
72:38 to that business who are not qualified, then they obviously don't want to be
72:41 taking calls or or doing anything further with them. So this process of
72:44 researching a new lead and deciding whether or not to take a call is known
72:48 as qualification, which is what this agent aims to automate. So the value
72:51 here is that instead of having to pay someone to manually qualify and go
72:54 through all of these leads or using arbitrary rules, which is what some
72:57 businesses have to go to, it's like look, oh look, we've got so many leads.
73:00 Let's just say if they don't want if they say that they're not on this, then
73:03 we'll just cut them out. And that's potentially leaving money on the table
73:05 by cutting out leads who would have actually been a good deal, but the rules
73:09 kind of didn't see enough detail to be able to determine if they're a good fit
73:12 or not. So, this automation is essentially immediately qualifying and
73:15 triggering the next steps to the sales team and allowing them to do that human
73:19 style research on these leads at scale. So, enough talking. Here's a little bit
73:22 more information. Again, with all of these, it's going to be on the figure on
73:24 the school. Then, I've also broken down how this agent would actually operate in
73:28 the real world and sort of real time. A lead's going to be submitting the form,
73:32 which is going to be this here. um the relevance company AI researcher. So
73:34 something that we've just built in relevance, we're going to reuse in here,
73:37 which is handy that you can start to move these components around and see how
73:39 they can fit into different automation platforms. Um then the AI agent is going
73:43 to look at this information from the research. Then it's going to determine
73:46 based on a qualification criteria we give them um inside the prompt of this
73:51 agent whether they are qualified or not. If they are qualified, it's going to uh
73:55 use this tool here and call this second workflow. when we're going to
73:58 essentially analyze that further and do a notification to either our agency team
74:03 or our our SAS team and if they are not qualified we will use this tool here
74:06 which will just send an email straight back to the person who submitted the
74:09 form say hey sorry we're not open to working with businesses like yourself at
74:12 the moment let us know if we can help you any other way so that's a rough
74:15 rundown of the build let's jump into it so to kick things off of course you need
74:20 a platform on natn.io io. All links and resources will of course be on the school uh post for this
74:24 video. And once you're on this page, you can go to get started and you can create
74:29 an account for free and just go through the setup process that they do. I'm sure
74:32 you can figure that out. They do have a 14-day free trial as of this filming. So
74:35 that's very good if you're just jumping in, not having to pay anything. And they
74:37 give you quite a lot of usage up here. As you can see, 1,000 executions. So
74:40 what we're going to do, of course, is click on create workflow up here. I will
74:43 be giving you the template. So if you want to just import it, you can. That
74:46 will be on the Figma there. But just like the last tutorial, I'm actually
74:49 going to be showing you the process of building these up from scratch that you
74:52 can see how I how you go through the process of building these automations
74:56 and the the testing and back and forth you need to do in order to get to the
74:58 end result. That's probably actually a lot more important if you are to go out
75:01 if I'm trying to teach you to fish, not just give you a fish, is to see how I
75:04 deal with problems when we're building these. So, let's get started by starting
75:09 off our trigger. We're going to go form has a nice form on new inn form event. And here we
75:17 get to create an N8N form. So we can call it uh work with your. So now we get to pick the field
75:31 names. So we want to have the first one is make that a required field. We add
75:40 another one. What is your company website? field. Right. So, we've just built out a
75:55 basic form here with first name, company website, which we're going to need for
75:58 the next step. I've put a placeholder in here so they know that it needs to have
76:02 https um at the front of it. And I'm asking for them to provide some
76:05 information about your inquiry like and maybe you can say what can we help
76:10 you? Um and that's going to be text area. So, they've got a bit more room.
76:12 So, we can go test step here. Make sure that it's all looking nice. This is what
76:15 the form's going to look like. What's Obby. There we go. We've submitted that
76:24 form. If you go back, there we go. We have the data in. So, this shows you the
76:28 output. NAM works by having kind of this middle island here, which is what you
76:31 set up. And then the left side is the input and the right side is the output.
76:34 So, we've tested it and these are the outputs that our form is giving, which
76:38 is what we are looking for. And the next step in this lead qualification process
76:42 is to do some research on the lead. So we have the company URL and this is when
76:46 we're going to go and make an HTTP request. So this is basically calling
76:50 any API over the internet. And in order to set this up, we actually need to go
76:57 relevance and we're going to find the company researcher tool that we made
76:59 with relevance. And now you're going to start to see how this all fits together.
77:03 um that building tools and relevance can also be very useful and an extremely
77:07 useful skill in all areas of AI automation because now I can come on to
77:11 research company and not only can I use it here this I mean this is why I think
77:15 relevance is such a great platform um you have the use here so I can send this
77:18 across to a client I can share it with it as I showed before I can use it just
77:22 here myself I can run it in bulk on a spreadsheet or but more importantly in
77:27 this case I can go to the API and now I can call this might look scary just
77:31 don't worry I can call this functionality basically send in a
77:35 company URL and get back the research. I can access this over the internet
77:38 through an API and they give me it here and they tell me exactly how to call it.
77:41 So now we this is on actually a post request. So we can copy this. Remember
77:46 how we talked about get and post request. This is a post request because
77:49 we're posting some data to relevant AI. So we change this method to post. We put
77:54 the URL in here. We do not need authentication in this case but you can
77:57 turn it on. So you can make a private here and then you can have an
77:59 authenticated. might sound a bit complicated but for now don't worry
78:02 about it we don't need to have an authentication step on this and then if we look at the request
78:09 body here it tells us how we can send data to relevant AI and if we go copy
78:16 here come back we're going to send a JSON and we can change this to using
78:26 JSON and then we can paste in basically what we have been given from relevant AI
78:29 now this looks a messy. Let's pop this open a bit more. And we actually need to change it
78:35 to an expression here. So fixed means that we're not accepting any dynamic
78:39 data in. So when the form is submitted, we actually need to take in some data
78:42 from that form which we have over here. We need to inject it into this uh HTTP
78:47 request to relevant AI to get this company research done. So we can't have
78:51 it as a fixed um JSON body here. Need to change it to expression and then we can
78:54 pop this out here and it gets a bit easier. So, we have params. We have the
78:59 company URL. And then we need to pop in here. Oh, the company
79:05 URL. Pop that in there. And here on the right side, we get to see what that
79:08 would look like given the test data that we've just put through. So, you can see
79:12 I've got the company URL, https morningside.ai in quotations here, which
79:15 is what we want. So, we can go back now. And now we can give it a test to see if
79:18 it's going to be able to communicate with relevant AI and get us the data we
79:24 want. There we go. We have our result back from relevant AI which is the
79:28 summary. As you can see when we go back to this um and back to build this is
79:33 exactly what we'd expect. You know you put in the URL does this scrape a fire
79:37 call writes the summary spits out the summary and this summary is what we're
79:40 getting back out over here which is what we want. So bank that's great another
79:43 step done. And those of you who are a bit confused about what this is this is just
79:48 the body of the request. So because we are sending a post request remember how
79:51 we have get and post request. Get requests are typically just with a URL
79:54 and with a bunch of stuff tacked on. A post request, we need to send a a JSON
79:59 body like this. And it does look quite confusing, but if I take this and go
80:09 form, paste this in and beautify it. And you can see how we have basically the
80:12 project which is the relevant project that we are calling. And this tells the
80:15 API this is the project that we want to interact with. and the project expects
80:20 the params, the inputs, which is the company URL, and we're injecting that
80:25 company URL from our information here in NA10 that we've dragged across. It might
80:28 seem tricky a few times, but trust me, this stuff becomes like riding a bike
80:31 once you get up and running. So, um, a few more of these and you'll be you'll
80:34 be completely fine. All right. And so now that we have our company
80:36 information, the next step is going to be setting up the agent, which is really
80:39 the coolest part in my opinion about make right now is that we come in here
80:44 and we can click on um agent and we can set it to as a tools agent, which means
80:48 we connect our own tools. And if we just back out of this, we can do cool things like set up
80:53 the model. And right here, this is so cool because we get to see exactly what
80:57 we were just talking about earlier in this video, but we have the different
81:00 parts of an agent, the different ingredients, right? So here's the chat
81:03 model. This is the brain. This is the LLM that's going to be powering the
81:06 whole agent using the soup example. Like this is the one meat that we get to
81:10 choose, right? This is a specific model. In this case, we going to be using
81:13 OpenAI again because you already have your API key and I can't really be
81:15 bothered going and showing you a whole another provider, but it's the same
81:18 process for all of them. You can if you want anthropic, you can then pick all
81:22 the anthropic models. Uh but in this case, just to keep it simple, let's just
81:30 to and open AI model. And if we go back again, you can see that we now have uh the memory and the
81:35 tools that we can connect. So, of course, we have the tools, which we've
81:38 talked about a lot in this video already. We can connect multiple
81:40 different tools here, as we're going to do in a second. And then we also have
81:43 memory set up here, which is a little bit outside of the scope of this video.
81:46 As I said, in most platforms, it comes builtin, but NAT is a bit more of a
81:49 developercentric platform. So, if you wanted to play around with memory, um
81:52 different forms of of managing memory, you can do that here. In our case, we're
81:55 not going to be touching it. And in order to connect a knowledge base, if we
81:58 wanted to set up a knowledge base for our agent, we would connect it as a
82:01 tool. So you can see here we've got in-memory vector store, pine cone vector
82:05 stores, etc. These are vector databases that we can connect just like in the
82:08 other tutorials we're going to do in this video. When you upload some
82:10 documents to make a knowledge base, they're essentially being put in a
82:13 vector store like these, but the platforms manage it for you and make it
82:16 a lot easier to do. So in this case, we're just going to be doing two
82:19 different tools. Firstly, we're going to be calling an NATM workflow. So I'm just
82:23 going to finish off the the basic setup of this. Um, and then we're going to add
82:26 another tool on. It's going to be the Gmail tool. And then before you know it, we
82:33 have our AI agent structure built out. So, we're using the Open AI models.
82:36 We're going to pick the model shortly. We're going to be using the tool to call
82:39 the second NATM workflow, which is going to be uh triggering the the email
82:43 notifications for our sales reps and the classification of the uh of the lead.
82:46 That's going to make a bit more sense when we actually do it. So, just stick
82:49 with me on this. And then this Gmail is going to be sending back a hey sorry you
82:55 didn't qualify for what we do um sorry we can't help you let us know if we can
82:58 do anything else for so to start setting things up we can start from left to
83:01 right here with the chat model um the openai model that we want to use and you
83:04 need to set up your openai account so you can click create new credential and
83:07 you need to go and add in your API key here I'd suggest you go and make a new
83:11 one on platform openai.com and you can add a new one in here and you can name
83:14 the key nat so you start to know which keys are used for which different
83:18 platforms um and once you put that in there you can Just click save and then
83:20 it will run a little test and then you're ready to go. You should have this
83:23 set up here. Then we get to select the model that we want to use for our agent
83:26 here. And in this case, I want to go for something quite smart. So, I'm going to
83:28 go for 03. We have 03 mini here. This one appears to be a little bit more recent.
83:34 So, they'll sometimes put the dates on the end of it. And if you just look at
83:36 the current date, you can kind of see how how close to the current date it is.
83:39 Um, but I'd say this one's a bit more recent, so it's going to be hopefully a
83:42 bit better. And next, we're going to skip this cuz we need to set up a whole
83:45 different workflow to connect it to. Um, we're going to jump straight into this
83:47 Gmail one. You need to set up again another connection as you go through all
83:50 of these different automation platforms. You do need to do these these
83:54 connections between uh your own account, say your Gmail or your calendar or all
83:57 these different apps. You need to go and create a new credential. Um and you can
84:01 just do the sign in with Google here. Super easy to do. I'm sure you don't
84:04 need my help with that. Once you've set up that connection, you can close this
84:07 and you will see the connection that you set up here. Basically, that's what
84:09 we're going to be sending emails through. And then we can get into
84:12 setting this up. So, because this Gmail tool is going to be used to send a a
84:15 reply back to the person who submitted the form and say, "Hey, sorry, you're
84:18 not a good fit for us." You want to send it back to the person who submitted the
84:21 form. Now, if you scroll down here, uh, yep, you see that I've I've forgotten to
84:25 add the email form in. So, this is a good example of needing to go back a
84:28 little bit. So, we can go back to the form submission here. Um, scroll down,
84:35 say, what is your email? And we can set it as an email. So, it's going to automatically force
84:41 them to provide a valid email for us. And I want to maybe shimmy this up a
84:49 Um, so it's name, email, then company, website. Um, we can do another test here
84:55 just to give it some proper data. Oh, it's not going to let us do
84:58 that because this isn't set up properly. So, we can just delete that for
85:03 now. And we're actually going to delete that. Otherwise, it's going to be a bit
85:06 of a pain. again. So, we can submit another form here and go back to NA10. And there we
85:20 go. We have the information. We have the email now. That's great. And so, now we
85:23 can come and set up our tools again. We've got the workflow there and we've
85:28 got the Gmail. Um, and now we have our Oh, and we haven't got the data here
85:41 because we need to run this again and get the research. So, now the
85:44 research is done in there for us to set up the Gmail tool here. We can go to two
85:49 and we'll be able to pull in the email. So, again, they submit this form. We
85:52 realize that they're not a qualified uh person for our offer. And then we're
85:55 going to send an email back and say, "Hey, sorry, you're not a good fit." So,
85:59 we can say the subject here is um thanks for your interest. I'm going to change
86:04 the email type to text here and I'm going to write a basic message in. I'm
86:07 just going to snag it from the one that I've done previously. So, we need to
86:10 change this from fix to an expression because we want to be pulling in their
86:13 name here. So, I'm going to paste in what I have here. Again, this would be
86:16 included um in the resources. It's just going to save us time if I don't have to
86:19 type this all out manually. Um but you can see here, I'll just delete this so
86:23 you guys know what we're doing. If I go hi, or hi, and then we can add in what
86:29 is your first name? Hi, name. In this case, it's going to be filling in my
86:32 name here as an example. So, highly mly, thank you for your interest in big boy
86:35 recruitment services as you specialize in recruitment for software and
86:38 development agencies. We're not a good fit based on your company's industry.
86:40 Please let us know if you'd like to connect us with one of your partners who
86:43 specializes in dealing with your needs. Cheers. Huge Jackman here to sales, big
86:48 boy recruits, BBR, uh, Dallas, Texas. So, if you guys remember huge Jackman,
86:52 comment down below um for the OG fans. And then that is our Gmail all set up.
86:56 And just quickly so that you've got a bit more knowledge around how this Gmail
86:59 uh tool works, we have all these different steps that we can use. We're
87:02 using the send one. So it's sending an email. You can use reply, you can use
87:05 get, delete, all these other functions, but the easiest one and the most common
87:08 one you're going to use is going to be that send one, of course. Now, we need
87:11 to set up this NATM workflow which the agent is going to call as a tool um when
87:15 they are a qualified prospect. So I'm going to delete this one and just save
87:18 this for now. Then we're going to go back to home. We're going to create a new
87:25 workflow. And this is a really cool skill that I want to teach you. The fact that you can
87:29 build all these workflows and then connect them to agents and it can just
87:32 be taking data in and kind of shooting it off in all directions and triggering
87:36 all these complex multi-step processes because it's a super valuable way of
87:39 using agents. Um, so I really want to teach you that. And obviously this one's
87:42 going to be starting off a bit different to the other one. We actually want this
87:45 to be set up as when executed by another workflow. So that's going to be what the
87:49 trigger is here. And we are going to be able to define using fields below. Let's
87:54 just add in one here that is a lead lead name. Um, for now we can just leave that
87:59 there. But that's all that we need to set up. We need to go back over to the
88:01 other one. Just needed to set this part up. Let's rename this qualified lead
88:07 lead classifier and notifier. So we can save that there. And
88:10 we have a bit more work to do on this other one. If we go back to this, we can
88:16 rename this here. So let's call this our lead qualification agent.
88:23 And so now we have our other workflow set up. We can come here. We can call
88:27 another workflow with the tool and we can call it lead is qualified. And so
88:31 now is when we get to tie back into what we learned in the foundation section
88:34 because we are now writing descriptions for our tools. Remember how we had
88:38 schemas and scheas are basically written instructions or instruction
88:42 manuals on how to use tools and how to use the APIs that wrap around them. Um
88:45 this description is going to be basically those descriptions that you
88:48 put in the schema. But NAT is going to be basically constructing it for us in
88:52 the back end. And we just get to put in here, okay, what's the name of this
88:55 tool? It's going to be called lead is qualified. It's giving us a nice example
88:58 of how we can write a description for the tool. So call this tool to get a
89:00 random color. The input should be a string with a comma separated names of
89:03 colors to execute. So in our case, we can say call this tool when the lead is
89:07 qualified according to our criteria. The inputs should be lead name, lead email,
89:18 company, company summary, and request in info. We're basically telling that AI
89:22 model or the brain what this tool can do. So when we send it some kind of
89:25 input, it then it looks over our tools. It looks at the Gmail description and it
89:29 looks at this uh workflow tool description and goes, "Oh, well, I have
89:31 a tool that does this and a tool that does this. What have they just sent me?
89:34 Okay, now I think I know what I need to do from here." So this is the rough gist
89:37 of what we want to do as a description for this tool. I'm actually going to
89:40 beef it up with a bit of a a bigger one here. Um if the lead is qualified to
89:44 work with big boy recruits, eg they are software based business like SAS or
89:47 development agencies and trigger this tool and send the lead data in the
89:50 following format. It's just dummy data. So name a name email um an email message
89:57 I want new div qualified true company information and company information
90:01 which is a summary of the relevance tool to do the company research that we have.
90:04 So, might seem a bit crazy at the moment, but stick with me because it
90:07 will make sense in a second. We're basically just told it when it's going
90:09 to trigger this tool and the format to send the data in. And then for the
90:13 workflow, we get to choose here the one that we just set up, which is our
90:17 qualified lead classifier and notifier, at least in my case. And then we see the
90:20 workflow inputs that we've just set up. So, if we go back over to our other
90:24 automation, so when we open this up and define our inputs here, you can see over
90:28 here we are getting just one of them that we've put in as an example so far.
90:32 So now we need to set up all these inputs correctly. And we have the name
90:35 that we want. So we've got the name and message. Um, honestly don't think we need this
90:54 qualified one here. And then we have the So, if we just test this, head back over
91:13 we refresh this list. Oh, back out. Save it. And this pops up. It says that these
91:21 inputs are outdated. So, there we go. We have lead name, email, message, etc. And
91:24 then we can actually automatically fill out a lot of these inputs. Maybe I will
91:27 put this back in here just to show you qualified. Um and then we just call it
91:39 uh true And we go back here and we add in one more which is a
91:58 qualified which is also a string. And so we have this qualified field here as
92:02 well. If we go back um I'll just test this. Save it again. Come back over and update
92:09 this. I think we can actually make it even cooler. So let's go. Um it changes
92:13 from a string to a boolean. So that's either true or false. Um, and if we test
92:20 this, save it again, and we change this to take away these little things. Sorry,
92:24 pisses me off if I don't have this set up right. Um, and then we update this,
92:28 you'll see this turns into a switch. So, that's a true or false, right? In order
92:31 to trigger this tool, it should be qualified by default. So, it's a little
92:34 bit redundant, but it's still cool to show you how we can get AI to fill out
92:37 these fields. Um, in a lot of cases, we don't, but for this qualified run, we
92:41 can. So, we can set this as an AI generated field. We can say if the elite
92:47 is qualified based on our criteria this set to true. And then for the rest of
93:07 quickly. Give this a second. Open this back up. And we can fill out some of these
93:14 fields. So we can go lead name that we want to pass through to the other
93:17 automation is going to be that. So we can fill a lot of these in email
93:26 um message and the company information um we can get from um this technically
93:30 but um maybe we could do a cool AI generated one here which is um a short
93:37 summary of the company and the industry that they are in. company's details. So, this is basically
93:46 telling the AI model of our agent how to fill this field out, which is one of the
93:49 reasons that AI agents are so powerful. So, that's all set up. Now, we've got
93:52 all of our tools set up. The last step is just to set up a prompt for our
93:56 agent. And I am going to cheat a little bit here and just throw one in that I've
93:59 done before to save us a bit of time here. So, we want to set the prompt
94:05 here. We can paste in this information here. And this is basically just telling
94:08 the agent who it is and what it's supposed to do. You're a lead
94:11 qualification agent. Your job is to analyze the form submission and company
94:14 research provided and then decide whether they are qualified to work with
94:18 big boy recruits. Ra we specialize in XYZ. Um we are specialist in capturing
94:22 talent for ra. We only work with softwarebased businesses, EG SAS
94:25 companies or development agencies. These companies are willing to pay much more
94:29 developers than your average marketing company or local business. Therefore, we
94:31 only work with them. Your job is to determine if the lead you are provided
94:35 with is a good fit for big boy recruits. And if so, call the lead is qualified
94:38 tool and send the elite information to it. If lead is not qualified, then you
94:41 must trigger the Gmail send email tool for us to respond to them letting know
94:44 letting them know we are unable to work with them. And then we have a response
94:47 format here which we can probably just delete. And then we can add in here is
94:50 the lead to information for you to analyze. Let's pop this out to make it a
95:07 We can add in um just go name um company URL company and we take it. So that's from
95:22 the relevant step for the research that we did to scrape using our firecrawl
95:25 tool. And then we provided all the information to this agent and it's going
95:29 to be injected with all of these values on each form submission and then it's
95:32 going to make a call on what tool it needs to use. So we're pretty much
95:37 there. We can even give it a run here and try to test the step and see which
95:46 choose. If we go back we can see okay look it's used the chat model as the
95:51 brain and it's triggered the NA10 workflow as expected. You can see here
95:54 that it's sent off information to our other workflow. It sent the lead name,
95:59 the email, the message, and the company information, and it set it as qualified
96:02 as well. So, all of these fields have been filled out. We've got a nice AI
96:05 generated summary here from the model and brain. And we have the qualified set
96:09 to true. And so, the final step now for us is to head back over to our other
96:14 automation and just finish it off. Oh, we need to save that. I will
96:19 just run that again for you so you can see it in slow motion. It's using the
96:22 LLM as the brain and the tools agent and it's deciding whether it's qualified or
96:26 not. And if it is qualified, then it will send it to this workflow. Bam,
96:31 we've sent it. And there you go. If we this. Head back over to our qualified
96:40 lead classifier notifier. Now, we can add on a quick few steps here. I'm just
96:43 sort of going to rip through this. Um it's not super important. Um but it just
96:46 shows you a little bit more functionality of what you can build in
96:49 on N10. So, we're going to add in a messenger model step here, and we're
96:57 mini. And what I want this to do is to take in that information that we sent to
97:00 the workflow about the company research, etc. So, we know this is a qualified
97:03 lead now, but we just want to split it between either our SAS team or our
97:07 development agency sales team. So, they're specialized in dealing with
97:09 different cases. So, I'm going to cheat and just throw in a prompt here, which
97:13 you guys will be able to get access to. Um, which is basically saying we have a
97:17 new inbound lead. Um, change this to an expression. Sorry. Um, we have a new
97:20 inbound lead that we need you to classify into either SAS or development
97:24 agency. Here's the lead information. Um, we need to go back step and test
97:30 this. There we go. We should have some information. Um, and now we can put
97:34 these in. So, you see how there was nothing here before I went back and
97:36 tested the trigger so that it gives us some null values here that we can fill
97:41 out. Here's lead information. Um lead name uh message um name
97:51 request company information if the company is a SAS output SAS if the lead has development
97:57 agency upput agency. So we're looking for just agency or SAS as the outputs
98:04 here. Um simplify the output. Yep, that's all good. So we can no point in
98:07 us testing that step there because all the values are null. Um but the next
98:12 step is a basic router flow if so this is a basic conditional
98:19 routing. So we have the conditions we can go expression here. So we can go
98:24 um the content here. So this is the output from the open AI step. If the
98:29 content which is the response from the LLM step the classifier it's either
98:32 going to be agency or it's going to be SAS. So if it let's just to to make it a
98:36 bit more flexible. If we go string, if agency, great. So, if it contains agency
98:46 on the true side, we want to go Gmail and we want to send a message. Um,
98:54 and then if it is false, we want to do basically the same thing. Now, I've got
99:11 Um, okay. So, here we're not getting much data on the input side here and we
99:14 can't seem to simulate it because it's of course triggered by another workflow.
99:18 What we can do is just save it here. go into executions. And if we go back to our
99:29 agent, and we can go to the form submissions one. If we go to executions, and we just
99:36 run one of these that we just did before, copy to again. This is basically just going to
99:48 trigger this again. and so that we get a fresh execution and we can sort of pull
99:52 that data back into the workflow. Oh, again. Boom. Triggered it. And that's
100:28 all done. Now if we go back to this and we go to executions, go to the most
100:32 recent one that succeeded. It's going to load in. Oh, hang on. This one's
100:51 it. Okay, so this one here, if we click this, yep, we've got all the data in
100:55 here. So, what we can do is copy this into the editor and then we've got the
100:58 data that we need that's already loaded in so that we have some values to put
101:02 into our Gmail steps Gmail. So, that's a handy little trick to to know how to do.
101:07 And now we have all of this information. So, that's what that's what I was trying
101:11 to get. Um, the same setup and we're going to send this to I'm just going to
101:15 use an example here and call this um it's the same email. You wouldn't pull
101:18 this in necessarily. I'm just using this as something that I can show you
101:24 at show. Say new agency lead. Let's do a text. We say um new agency lead man. Go
101:30 get him. Um turn into an expression. And then we say we can just
101:37 throw this company data in there. It's going to be messy. You can play around
101:40 with this more when it comes to formatting, but just to show you the
101:44 functionality. If we go uh test a step here, that's going to sent an email to
101:49 this. This is like my agency sales reps uh email. Of this. Can duplicate this. Right click,
102:06 duplicate, bring it here. Oh, connect this up. And we change this.
102:14 You change this to your like SAS guy. You change it to a different different
102:18 email. Um, of course, and then you can say new SAS lead. Right. So now we have done all of
102:27 that basically all built out. The data is going to come in from the agent. It's
102:30 going to send in the company summary. This is going to classify it into being
102:34 a uh agency lead or a SAS lead because those are the only two types of
102:37 businesses that we work with. So all of them will be qualified when they come
102:40 through here. And then it just sends an email to our uh agency sales rep or our
102:46 uh SAS oh rename this to our SAS sales rep, new SAS lead for them to continue
102:56 with. Right. So to test this we can turn this on to active and you can see that
102:59 you can now make calls from your production form URL. Um we can go okay.
103:05 If we double click on this we can open this up. We can click on production URL.
103:11 Copy this and open this up in a new tab. spin. So of course my agency Morning
103:20 Side AI does development services. So this should be qualified and it should
103:24 also route it to the agency email. So if I now go submit, we go back into NADN, we go into
103:32 executions, we can see this one is running. If we go to inbox and there we go. We see it has
103:47 succeeded. And then if we go to and then if we go to our lead
103:51 qualifier and notifier and we go to executions, we will also see that we
103:55 have a new one that has succeeded here which was just uh a few seconds ago and
104:01 that's gone through. It is um outputed it as agency which is the the
104:05 classification that we wanted. It has gone through and has sent a new email.
104:08 And if we go to here we have new agency lead there. There we go. All the
104:11 information. So that's working number one. Now we can go back to our form and
104:14 we can try it again but this time with let's say an unqualified business. Let's
104:22 go. What is your name? Ray Croc Ray McTum I need more guys more people
104:33 flipping damn burgers. So essentially Ray here has come to our recruitment agency and
104:39 they're asking, "Hey, I need people to do flip patties for me in my fast food
104:43 restaurant." Um, and because Big Boy Recruits in Dallas, which is a
104:47 hypothetical company, of course, um, doesn't do that. It's going to qualify
104:50 them or it's going to disqualify them and then send an email to our good
104:54 buddy. Oh no, some poor dude at McDonald's is going to get an email now. Um, because
105:00 send an email. Um, but it's going to be running and of course it's going to be
105:04 sending. if they're unqualified, it will send an email to them and say, "Hey,
105:07 sorry you're not a good fit for us. Let us know if we can help or we can connect
105:13 partners." And while that is running, I would just put together the final one
105:16 here to test the functionality, which is if we go Liam admin.com, and we set up my SAS
105:24 https, my SAS agent, if you haven't already used it, we're going to show you
105:26 how to use it in the last tutorial of this video. So, you guys will get to see
105:30 that. Um, which is my own no code AI agent building platform. And what can we
105:34 help you with agents? Um, so this should be a SAS one and it's going to qualify
105:45 have the McDonald's one has succeeded here. And you see, yep, as expected, we
105:49 were not qualified. The McDonald's person was not qualified for our offer.
105:52 So, it looked at the qualification criteria we provided in here, said,
105:55 "Hey, no, that's not a good fit." So, I'm going to use this tool. And you can
106:00 see that it sent the email and it said, "Hey, thanks for your interest. Um, but
106:04 we're not a good fit for you." So, someone at McDonald's just got an email.
106:07 Apologize for that, but we didn't trigger the other workflow, which is a
106:10 key part. And we're not going to send emails to our sales team saying, "Hey,
106:14 look, new leads." Now, I have sent another one through here which just
106:17 finished executing. And we can see this. It's gone through. It's researched um
106:24 agentive. you'll see um Agent is a leading service delivery platform for AI
106:29 agent AI automation agency owners um etc and it's called the tool because we were
106:32 qualified because we're a SAS business right and again if we go back to
106:38 here and we look at the most recent one here then you'll see new SAS lead has
106:48 been triggered because we are a SAS of course um the LLM step here has outputed
106:53 just SAS So that means that it should send an email to the SAS team, which if
107:02 inbox, tada, new SAS lead, right? So I know that may have taken a while,
107:06 but uh we got there eventually. And you can see that we've built out all of this
107:09 functionality. We have our AI agent calling our tools if they are qualified
107:12 and triggering this other workflow. Again, you can build so much cool stuff
107:15 by connecting an agent to multiple different workflows. We have a little
107:18 relevance AI researcher tool that we're reusing here and we have people getting
107:22 denied um with an instant email sending them back. So, hope this been a cool one
107:26 to show you how NATM works. I really, really like this agent functionality
107:29 that they have. I think you guys are going to be able to build some awesome
107:31 stuff if you keep going down this rabbit hole. So, that has been agent build
107:35 number two. Stick with me as we jump into agent build number three, which is
107:39 a pretty damn cool one, focusing on both chat and voice-based agents all in the
107:43 same build. So, let's get the ball All right, so that is two builds out of
107:51 the way. Well done if you made it this far. We have another big one here. Um,
107:55 this is going to be breaking down how to use voice flow. Let's take this off
107:59 here. Um, to build an agent that is going to be both accessible through a
108:02 website chat, so you can chat to it on a website as a as a chat widget, which
108:05 you're probably familiar with, but we're going to connect the exact same agent
108:08 and exact same functionality that you get through that chat widget also to a
108:12 phone number on the website that will be able to call and have the same
108:14 experience. So, this is going to show you how on voice you can build both chat
108:19 and voice experiences. Um, and this is a new feature for them as of the time of
108:21 filming. And this agent is what we can classify as an AI customer support and
108:25 lead generation agent for both website and phone. And we're going to build it
108:28 on voice flow, of course. And the purpose of this agent is that it's
108:30 designed to be able to answer common questions from potential customers via a
108:33 website chatbot and also via a phone number that can be called. Not only can
108:37 it answer questions to help them sort of move them towards a uh a purchase, but
108:41 it can also generate instant quotes for interested parties. This is intended to
108:44 increase the number of leads that they get because people who see a contact
108:47 form may be like, "Hey, I want to get instant response. I want to know
108:50 instantly how much this is cuz I'm shopping around." Um, and rather than
108:53 just filling in a contact form and waiting. Um, having this instant
108:56 quotation can give people confirmation on the price. Um, so they're ready to
108:59 take a step forward and end up getting the sale ultimately. So that instant
109:02 quotation feature is a cool one. Um, very easy to do with the custom tool on
109:05 relevance that we're going to build. And finally, this agent is going to be able
109:08 to actually capture lead information from those who have been given a quote.
109:11 So after they've been given a quote, then we'll move to say, "Hey, give us
109:15 your details and we will follow up. Our team will follow you up and set an
109:17 appointment for the service." And the value here of the system is that
109:20 customers are going to often want instant answers so that they can make a
109:23 purchase. So by offering easy ways for them to get this information, we can
109:25 increase the chance that they're going to purchase from the business. Um,
109:28 companies typically have to spend money on some kind of customer support or
109:32 sales team in order to get these kind of answers given to customers when they
109:35 need them. But this agent can essentially be a oneanddone solution um
109:39 to both help increase the sales of the business by increasing that likelihood
109:42 of purchasing because they now have more information um while also saving the
109:45 business money that they would typically spend on some kind of support staff um
109:50 say if this chatbot can handle a dozens and dozens of responses a a week that
109:53 would typically have to have gone through a support person then we're
109:56 saving the company money and also helping them increase their chance of
109:59 generating more revenue. So um here's a rough layout of the design here. We are
110:02 going to have a website. I'll give you a template for this. It's very easy to set
110:05 up and we're just going to throw in a number um that's going to be connected
110:08 to the voice agent that we build and we're going to be setting up voice flows
110:11 web chat widget as well. And this agent is going to have access to a knowledge
110:15 base to answer questions um that prospects may have about the business
110:18 and their services etc. Um it's going to have a tool that is allowing them to
110:21 generate an instant quotation. So it's going to take in some information. This
110:24 is going to be for a cleaning business or a hypothetical cleaning business. And
110:26 then we're going to be able to generate an instant quote for them based on the
110:29 property type and the size of the property they need cleaned. and we're
110:32 going to be able to capture the lead information afterwards and log it into a
110:35 CRM. In this case, we'll just use Google Sheets, but it's fairly easy to swap
110:38 that out to whatever CRM you want. So, it's going to look a bit like this. I've
110:41 actually added a little bit more. And we are using another relevance tool in
110:44 here. This is from a different project from my accelerator, but I'm going to be
110:46 pinching that and putting it in here for you all. And this is the uh tool number
110:51 two here, the generate instant quote. So, we're going to be slotting that in
110:53 there, taking in some information, answering questions, etc. The process of
110:56 building on Voice Flow is one of my kind of favorite experiences um in the
110:59 automation space. I really like the the way they've built out their uh their
111:02 flow builder. Um so I'm sure you guys will enjoy building this uh step by step
111:05 with me. And then the general usage pattern of the system is that the
111:08 person's going to arrive on the website. They'll either click to chat with the
111:11 chatbot and engage with this functionality or they'll enter the phone
111:13 number into their phone. And then the agent will jump in and respond either
111:17 through text or through uh voice and determine what they're needing help
111:19 with, which is this section here. And then it's going to be routing using this
111:23 router section here to the correct tool whether they want a question answered or
111:27 they want to get a quote. Um, and then each of these branches will execute on
111:30 that uh, functionality depending on their intent. So, it's going to look a
111:33 bit like this. We'll have a phone number and we'll have a chatbot like this. This
111:36 is actually an agenda chatbot from my own software, but we'll be swapping this
111:39 out to a voice flow one in this build. So, without further ado, let's jump into
111:42 voice flow. So, when you click that link on the Figma, it's going to take you to the
111:47 signup page. You can sign up there and then once you're in, you're going to get
111:49 a page that looks a bit like this. The first thing that we want to do, of
111:52 course, is to create a new agent up here on the top right. Let's call this
111:56 Bonor's cleaning um, website. and phone agent. Um, let's just start with a basic
112:00 template here. Import knowledge for this import knowledge. We can actually just
112:03 skip that for now. And then we get into the flow builder on voice flow. So just
112:07 a quick orientation if you are new to the voice flow platform. This is where
112:10 we can add in our knowledge which we will do shortly. The workflows are where
112:13 we access the flow builder. In most simple builds like this, you just going
112:16 to have one workflow. So you don't need to worry too much about that. Now we
112:20 have integrations like uh the widget which we're going to be using to deploy
112:23 this on a website. We have the phone number integration which we're going to
112:26 be doing later as well. Then we have API keys etc which you don't need to worry
112:29 too much about right now. We have some publishing features here which we'll
112:32 double later. We have access to transcripts. So once we deploy this you
112:35 can access all of the transcripts either by voice or through chat here and and
112:39 sort of dig through the answers and and see how the uh people are interacting
112:42 with the agent that you've built. Um something that a lot of people neglect
112:44 after they've put one of these into production. And then we have things like
112:48 analytics um etc. But obviously we need some data before we see anything useful
112:50 there. And then the settings page is not too much you need to worry about right
112:53 now. Just sort of on a need to know basis. The more important stuff, of
112:55 course, is up in this first tab here, which is content. So, we have messages,
112:59 we have prompts, we have components. We're going to be working a lot with
113:02 prompts shortly. So, that's the main one we need to be worried about. But for
113:05 now, we can just go into workflows, and we open up this first workflow and edit
113:09 it. And here we have the template that VoiceFlow gives for us. Um, which I'm
113:12 actually just going to nuke this, and we'll start fresh. And if you see on the
113:15 Figma, we have a design here that we're roughly working towards. I'm going to be
113:18 showing you the sort of step by step. So, we need a welcome step.
113:22 So, we're going to start off by going here and dropping. I'll try to zoom in a
113:27 bit for you guys here. Talk message. So, message is how you send a message um via
113:31 the chat. So, start is when maybe you click on the widget, it pops up. And
113:34 this message that we're about to put in is the first message the bot is going to
113:39 send. So, we say um up here, hey, hey, welcome to corners. I'm going to zoom this up for a
113:44 bit for you guys. And right away, we have a little tip and trick that I want
113:47 to give to you because we are building this as a chat and voice assistant. We
113:51 want to over voice. You don't want to overuse punctuation because it leaves
113:54 these big long pauses when the the voice agent is going, "Hey, welcome to
113:59 Connor's cleaning." So, we wanted to just say, "Hey, welcome to Connor's
114:01 cleaning." A bit more natural. There's times where you'll see me on the side of
114:04 sloppier punctuation, but that's just to ensure that when we get to testing it on
114:08 voice, it sounds as natural as possible. So, hey, welcome to Connor's cleaning.
114:10 And then, of course, we're going to wait for them to reply and say something back
114:14 to us. We go to listen and then capture. And this is going to capture the
114:17 information. We want to change this from capturing entities which is like say I'm
114:21 looking for a price or an address. Um we're just going to go the entire user
114:24 reply and the reply is going to be saved into this variable here. So we actually
114:30 want to change this to um first user reply because we're going to need it a
114:37 little bit later. Um the users first reply. And I like to name these as we
114:41 go. So we can call this welcome. Drag this out here. Get a new
114:45 one. And we're going to be doing a uh a set step here. So, we're going to be
114:48 setting some variables. And I'm going to add a new variable to set. And we're
114:52 going to do it based off a prompt here, which is a cool feature in voice they've
114:55 added recently. And we're going to be able to select a prompt that's going to
114:58 take this information from this first reply and then generate some kind of
115:01 output from it and set a variable. And the variable we're going to set here is
115:04 called last response. So, this is typically what you're going to put as
115:06 the last response from the AI or from the agent. Um, last response here. And
115:10 last utterance is typically the most recent uh message from the user. So
115:13 utterance is coming from the user the most recent last utterance and the last
115:18 response is what the the AI or the agent or the system has last responded to. So
115:23 we want to set the last response to something that is generated through this
115:25 prompt here. So we can create a new prompt here and this is basically going
115:28 to take in the data from the chat and the conversation so far and we'll be
115:31 able to generate things off of it. So say we add in here the conversation
115:35 history. That's a good thing to have in in most cases. And I'm going to be
115:37 dumping in some of the prompts here just to save us a bit of time. But basically,
115:40 we're saying summarize the customer's question below and ask them to confirm
115:43 that that's what they meant. And so, we're not actually going to be
115:45 generating the last utterance here. We're going to be adding in the last the
115:50 first user reply that we got. I mean, it's going to be included here in the
115:53 conversation history, but there's no harm really in hard coding it or at
115:56 least putting the variable in here to make sure that it's in there. Um, we're
115:59 just saying summarize the customer's question and basically say a
116:01 confirmation statement. So, just to confirm this is what you're looking to
116:04 do. So, you imagine this over the phone. Hey, um um yeah, I'm not really sure
116:08 what I'm supposed to be doing here, but I was thinking if if you guys were
116:12 possibly so all of that information is taken in, we can flick back to them and
116:15 say, just to confirm this, this sounds like you're looking for this, yes or no.
116:18 Um so, ensure your tone is empathetic. Speak directly to the end customer. Keep
116:22 your answer brief and two sentences max. So, if we go back here and actually we
116:28 can name that prompt. So it's um summarize problem and then we need to send the
116:35 response. So this prompt is going to take in the information we provided
116:38 here. It's going to use this prompt to take in um the conversation history so
116:43 far and this information from the user in the first question. It's going to
116:47 generate a a question to ask back and it's going to save it to this variable
116:51 that we have here. So apply output to variable last response. There is
116:54 actually an easier way to do AI responses like this, but in our case, we
116:57 need to be saving this variable. So, it'll make sense in a second. But, we
117:01 can go into here and we can go last response. And then it's going to send
117:05 that information back to them. So, let's just do a quick test here. Click start.
117:11 Hey, welcome to Connor's Cleaning. Oh, actually, we need to ask a question.
117:15 Hey, welcome to Connor's Cleaning. Um, We'll say I need cleaning services for my
117:30 house. Sounds like you're looking for cleaning services for your greenhouse.
117:33 Is that correct? Want to make sure I understand your specific needs before we
117:36 proceed. So, I obviously spelled house with G house. So, we thought it was a
117:39 greenhouse, but that's what we want. Some kind of confirmation message just
117:41 saying like, hey, look, is this what you're actually looking to do before we
117:44 then go and trigger the different tools that we are equipping our agent with? By
117:47 the way, there is a way of changing between trackpad and mouse. So I am
117:50 panning around with my mouse here. You can also do a trackpad method which is a
117:53 lot easier to use if you're if you're having trouble using it. Okay. So after
117:56 that they're obviously going to say yes or no whether like have I got the
117:59 question or have I summarized what you're looking for correctly and we can
118:03 go to a choice step here and we can set up some triggers. We can set the intent
118:07 to yes and then we can add another trigger and we can set it as no. So this
118:12 is basically using AI to analyze what they've said and grouping it around
118:15 these certain things which are called intents. So, what is the intent of them
118:19 of this uh of the response? And in this case, they have some pre-built ones, but
118:21 we are going to be building our own custom intents later. But for now, just
118:25 know that if you're looking to sort of split traffic or split people coming
118:28 through the system, these choice blocks with the default uh intents from voice
118:32 flow, yes and no, are ready to use out of the box. And if they don't say yes or
118:35 no clearly or we can't pick it up, we can add a no match here. We can say
118:39 sorry, I didn't get that. Can you say yes again? A yes or a no is enough. And
118:45 then we can say to follow a path after these reprompts which we'll call no
118:49 match. And then we have this uh no match path which we'll set up in a second
118:53 here. I'll just put it as a a filler for now. Basically if if they don't say yes
118:56 or no um this is setting up error handling. Um, and basically if people
119:01 particularly over the phone, um, there's so many different ways that the
119:03 conversation can go and end up and you'll want to, while I don't focus on
119:07 it too much in this build, um, as you're building production grade assistants,
119:10 you'll need to build a lot more of these fallbacks and these reprompt and these
119:14 no match things to handle edge cases where people use it in a weird way that
119:17 you don't expect. So, I want to give you a little taste of that in this tutorial,
119:20 but it is nowhere near representative of what it takes to actually get something
119:23 into production that you can trust on a on a customer's website. Okay, so we've
119:26 got this choice block set up to determine if we have got their
119:29 summarization of the problem correct. And we can take this up to here and we
119:33 can set another uh variable. And so this is really the core part of the
119:36 application which is determining what their intent is, what are they looking
119:39 to do and which tool are we going to route them to it. So this is a very
119:41 handy skill to have which I'm going to teach you which is how to set up some
119:46 kind of intent classification system. Um which is really really essential to
119:49 building agents on on voice flow and any kind of agent where the platform itself
119:52 isn't automatically handling that for you. So if we go set a new variable and
119:56 we're going to do it through a prompt. We're going to set a new variable here
120:00 called desired action. So basically people coming through and asking questions can
120:07 be saying hey look I I just have a quick question about where you guys are based
120:09 and then we're going to route them to the knowledge base. And then someone may
120:11 be saying hey how much does it cost for this? And then we're going to route them
120:15 to the pricing uh the instant quotation um system that we're setting up. So
120:17 needs to be able to determine what they're looking for and we're going to
120:20 route them depending on that. And that's what this router is going to do. So the
120:25 action that the prospect wants to take, the most likely action that the prospect
120:29 wants to take. Then we need to make a new prompt here. We're going to call
120:34 this intent classifier. Classifies the intent of the uh prospect into asking the knowledge
120:44 base or generating an instant quote. Add in the conversation history.
120:47 It's always good to have that in there. And I'm going to put in the prompt that
120:49 I've written previously. And this is a pretty basic one as well, which is just
120:52 saying what does the customer want to do, ask a question, get a real-time
120:55 quote, or something else entirely. You must output a label for this only. Your
121:00 options are ask a question, get a quote, or other. And you guys can just pause it
121:02 and see what I've got in there. But basically, anything asking a question is
121:05 going to be about the business and the services. And if there's anything about
121:09 pricing or directly related to getting a quote and like they're ready to move on
121:12 this, then we're going to route them to get a quote. And anything else is going
121:14 to go to other. And because in the next step, we are going to be looking out for
121:18 either this is the output or this or this. a really clear statement saying
121:21 this is all you need to output just this and not hey I took a look at the the
121:25 conversation history and it seems like the user wants to do get a quote we just
121:28 want just get a quote and we can explicitly state that with this big caps
121:31 lock block here and as with the other builds all of these prompts are going to
121:34 be available in the resources for you to follow along with okay so now we get to
121:39 the cool bit which is routing this. So if we go condition add this in here and
121:46 we go add path condition builder and we say if desired action
121:51 is ask question. Oh that's that's all we need there. So as you can see that's added
122:02 one in here. And if we want to add another one in, we go if desired action
122:11 is um what do we have the label as output? What did we what's the exact
122:16 label that we had in here? So this prompt is going to be outputting these
122:19 labels. So we need to make sure they match up. So ask a question. Get a
122:29 Yep. Get a quote. other and it's already got an else path in there for any error handling as well.
122:39 So what this now allows us to do is to build out our different tools. So we can
122:43 go up to here to ask a question. Just throw this in for now so we can get an
122:47 idea of what it's going to look Um other this is going to be sort of
122:54 error handling. And if it's else, that means that the LLM step here hasn't outputed
123:00 any of the labels that we told it to, and it's likely thrown in a bunch of
123:04 rubbish. Um, so this is sort of an error handling step. Um, should say, "Sorry,
123:09 something went wrong, at least during this prototyping phase. So now what I'd
123:11 like to do is make this look a little bit prettier. Um, we can go through and
123:16 add things in here like this is the uh confirm problem um intent classifier
123:23 router." And then we can go here add a note can say tool number number one
123:37 base two you guys don't have this so be easy for you guys to see understand I
123:44 confusing and then this other one we don't need to worry about too much so to
123:46 keep things quick I'm not going to test this just yet I'm going to test it once
123:49 we've got that functionality set up on either side at least for this top one
123:52 first so now We need to go and set up our knowledge base. And to do this, we
123:55 can click on the back button here. Go to our knowledge. And here we have a data
124:00 source which we can upload. I'm going to upload a file, but you can put in URLs
124:04 to different websites, etc. I'm going to be uploading a file
124:08 here. And I'm going to upload this Connor's cleaning FAQ kind of document,
124:11 which you guys are going to have access to in the resources. Basically, it's
124:15 just about us, location, our services, ra um some frequently asked questions,
124:20 etc. So, I've just AI generated this. Um, and if you're doing any kind of
124:23 prototype builds, I recommend you do the same just to throw it in there and see
124:25 if the knowledge base is working as expected. Obviously, you'd swap this out
124:28 with actual customer data or or your client's FAQ. I'm just going to throw
124:32 that in there for now. And you guys can do the same. And when you're setting up your
124:38 knowledge base, you can also set up the settings for it. So, in this case, it's
124:42 using by default Claude 3.5 Haiku. And you can see how many tokens this is
124:45 going to cost you for Voice Flow's usage. Um, what is Haiku? Haiku seems to
124:49 be the cheapest. Oh, you've got GBT40 mini. Let me just chuck on GBT40 mini
124:53 here. We want this thing to be pretty deterministic. So, I'd say 0.1 is fine.
124:57 Max tokens. Um, we can increase that just in case it needs to give a longer
125:02 answer. And chunk limit of of three should be enough. So, that's just so
125:04 this stuff is a little bit more advanced. That's that vector database as
125:07 I was talking about. Basically, knowledge base is going to be sending
125:11 the message that we ask it and querying it and getting back chunks of
125:14 information. Because our knowledge base is quite small, we don't really need to
125:16 have too many chunks. If you put this up, you just be getting the whole
125:19 document back. Anyway, so max tokens, the number of tokens that it's going to
125:22 include in the response. So, we want to increase that to 480 um so that it can
125:25 give a longer response if they need. Maybe just tone that down a bit. And
125:28 these are of course kind of controls that you have on how much you want to
125:31 allow the app to spend. And those settings obviously the main ways that
125:34 you can control how much um your knowledge base is using and how much
125:37 your you or your client are ultimately spending on the AI features for the
125:40 knowledge base. So once we've got that set up, we can go back to workflows
125:43 here, open this up again, and then to plug in our knowledge base, we can go to the dev section here.
125:49 We can go to KB search, pop this in here. I will uh we need to delete that and reconnect
125:57 this up to the top. And we're going to delete this as well. And so we can go
126:02 into this knowledgebased step here, and we can enter the query. So what we're
126:05 going to say is we basically want to throw the information that we've got
126:08 from the user already about what they want which is we have here as the first
126:12 message they gave us which might be a bit longwinded. Then we have the summary
126:15 that they have confirmed and then we can put these into the query that's going to
126:19 be asking the knowledge base hey this is what we want information on can you give
126:24 us some information back. So we can go user first message put a curable in
126:29 there. We can go first use reply and then we can also go summarized problem.
126:33 Now you can see why if we put last response here, why we have this variable
126:36 saved instead of just sending it automatically. So I actually don't like
126:39 using last response because that's something that you like to update quite
126:41 a lot. So I'm actually just going to switch this to um changing it to
126:49 summarized problem just so we don't get any kind of overlaps that cause problems
126:55 down the line. A summary of the user's Then we put this in here. Get to spit it
127:07 out. So when you put a variable in a message, it's just going to print out
127:10 and and spit out the the value that's inside that variable. So we've set the
127:13 summarized problem variable and then we're just going to spit it out and send
127:16 it into the chat or over the over the phone. So now we can come back out to
127:19 our knowledge base and we can take out the SL response and replace it with
127:23 summarized summarized problem. Then we can save the chunks that come back from
127:26 the knowledge base. I don't want to get too in depth on what chunks are
127:29 specifically. It's a little bit more advanced, but for now, we can just know
127:31 that it's going to return some information from that knowledge base.
127:34 It's going to chop up that document we put in. And when we put in this
127:38 question, it's going to basically ask that knowledge base, can we get three
127:41 chunks that most closely match the information that is in this query that
127:44 we sent to it. So, we can save these chunks, which is going to be three
127:46 because we set that up in the knowledge base settings into this chunks variable.
127:51 And the chunk limit is three still. If we click this, we can add in a chunks
127:53 not found path. But for this tutorial, we don't need to worry about that
127:56 necessarily. And then we're going to use those chunks that came back from the
127:58 knowledge base to generate an answer based on the original question. So if we
128:02 put this here, we go talk, we go prompt. And for this prompt, if we go here, we
128:06 can create a new prompt. We can add in the conversation history just for good
128:09 measure here to give it the full context of what's going on. And then I have a
128:12 prompt here. You are an AI customer support rep from Connor's Cleaning
128:15 helping customer with the question. Use the provider details below to answer the
128:18 customer's question. Ensure you keep your answer brief and speak directly to
128:21 your end customer. You are speaking to them over the phone. It's the input data
128:25 provider details which is the chunks variable. I'll just put that in
128:30 again. Chunks to make sure it's set up properly to the user's original uh
128:36 question. We can put all this information back in. So, first reply,
128:39 this is what they asked us as soon as they picked up the phone or the first
128:42 message they sent when we asked them what can we help you with. And then we
128:46 also just put in for good measure our summary of the problem that they
128:51 confirmed. And we can go to our summarize problem variable and throw that in. That should
128:56 be good to go. And I like to make these look purple or some kind of cool color. Um, call
129:04 this a KB query. And we can change this to generate answer from chunks. Let's say
129:13 from Kh. And if you want to make this a bit easier for you to kind of understand
129:17 at a glance, you can add in your descriptions on these. So if we go to
129:22 edit again and we go here, this takes in chunks from the from the KB and their
129:28 original question and writes a short and sweet answer. And we have this in here
129:31 that you are speaking to them over the phone because we want to make sure that
129:33 we're building this with the phone in mind, which is more tricky than just
129:37 chat. So long text outputs don't really work that well over the phone. So that's
129:39 kind of why we're putting that in there as well. And we can call this um
129:44 generate answer. All right. So now we can actually give this a spin. We can start
129:55 again. Oh, we may need to if we just click run. Okay, there's no training
129:59 needed yet. We can run test. Where are you located? Sounds like you're asking about
130:05 a business location. Could you confirm if you'd like to know the specific
130:08 address or where Connor's Cleaning operates? Okay, it's good this popped up
130:10 because as you can see, it's asking for a non- yes or no answer. We're just
130:15 looking for a confirmation in yes or no. And so this would technically break the
130:18 system and that'd be saying, "Oh, I'd like a specific address." And this is
130:21 looking for yes or no. And so it would send it to this no match. So what we can
130:24 do to fix this is to go into the summarize problem prompt, modify the
130:29 prompt, and then say they should be able to answer only with yes or no. This is a
130:35 confirmation step, not asking for more located? Sounds like you want to know
130:56 the specific location of our cleaning business. Is that correct?
131:02 Yep. Now it goes to the router here. It's going to determine that I said yes.
131:06 Bam, bam, bam. And there we go. So that that all happened pretty quickly, but
131:09 you can see it's sort of broken down through here. Um, if I click over here,
131:13 it's going to remove all of this. Okay, so let's break down how this happened
131:16 step by step. Um, so you can see this through here. It's still using 3.5 haiku
131:20 for some reason. I'll need to double check why that's still using the model
131:24 we didn't select. But basically, it comes through this step here is the
131:27 intent classifier. So you can see that it set it as yes. That is the correct
131:31 intent and predicted intent yes. And that routed it to this. And then using the model again, it
131:38 analyzed the information that it was given. And then it set the desired
131:42 action variable to ask a question which is one of the labels that we wanted and
131:45 that is correct. And then it said condition matched taken path one. So it
131:49 set the desired action variable to ask a question. We were checking for it here.
131:52 Then it said okay great. Now I'm sending it up here to the knowledgebased query.
131:57 It says it's query received. We passed in all the information whereabouts you
132:00 located and then the summary that we gave it. And then we got two chunks back
132:04 from the KB and the AI response here finally took in all of this information
132:08 and it gave us the final output and generated this response. We are located
132:12 at 247 ra and at the end here it's saying is there a specific area you're
132:16 interested in. For a basic build like this I'd probably change the prompt to
132:19 say don't ask another question because in this case you then need to set up a
132:22 looping mechanism where it can keep answering questions for them and then
132:25 break out into any of these other intents um as needed. But for now that
132:29 is a knowledge base and that's how you can ask questions. And so that is tool
132:32 number one knocked out which was easy enough. So great we can go on to tool
132:35 number two now. So for this second quote I'll be able to give you a relevance
132:38 tool that we're using for it. But let's just jump into answering this question.
132:41 So they've said here that they wanted something related to pricing or quoting.
132:45 Remember in here in the router we have set the intent classifier to say it's
132:48 going to go to the get a quote if they're asking about pricing or have
132:52 directly requests a quote. And this will take them to a real-time quotation tool
132:55 that takes the property type and size and then returns an estimate. So that's
132:58 the people that are going to be getting to this next branch of the agent. So
133:04 desired action is get a quote. We can say, "Okay, sure." To give you an
133:14 just Okay, sure. To give you an instant quote, I just need the properties type
133:18 and size and square feet. Then I'm going to add another chat step or message,
133:32 apartment? The next we can do one of these choice steps again and this time
133:36 we get to create some custom intents. So we can go to triggers here and we want
133:39 to select an intent. See it doesn't have a house. So we can go create an intent
133:51 users property type is a house. And then we can add in some examples here. So
133:54 obviously house this is just giving examples to voice flows AI engine to
133:58 help us better to classify the different intents as they come into this step. So
134:02 we can say home and then you can add in some AI generated ones here which
134:06 usually pretty easy and uh which usually pretty good. Okay, residence, dwelling,
134:10 property, abode, living place. Uh this this gets tricky um because some of
134:15 these could overlap with apartment. So property is probably abodess too broad
134:18 living place. uh dwelling uh residence potentially we could get away
134:26 with. We could say like home single family home and by now we've given enough
134:39 examples where we can just go to create and now we need to do the same for
134:43 apartment. So, we'll go to create an intent apartment. The user's property
134:49 type. The user's property type is an apartment. Um, see what else it's got
135:02 for us. Um, townhouse, penthouse, duplex, flat, probably not
135:08 right. Loft probably not right. Townhouse penthouse. Think that's a good
135:12 bunch. And then we can add in a no match here as well. Um, and we can add in a reprompt. Sorry,
135:29 an If it is a house, we can come up here set. We can go value or expression. Select the
135:42 variable to set. We're going to go property the save that variable or create that
135:53 variable, sorry. And then we can enter this in and set this as
136:00 house. So we're setting the property type variable to house when it's been
136:05 triggered um by this particular route. And then we need to add another
136:11 one. So, I'll probably just duplicate this. That's by right clicking on one of
136:15 these blocks. Um, and then we can connect up apartment to it. Property type. Instead of being
136:25 apartment. And from here, now that we know the property type, we can set the
136:29 size of the apartment or the house. So, we go and how many square and how many square feet is it?
136:39 We can connect both of these up here. And then we're going to save the entire
136:43 user reply or on a capture step. So anything that they respond to after
136:52 thing. And we want to go here, set up a set using a prompt. And what we're going to
137:10 do is use AI to analyze the response and then extract the number of square feet
137:14 from it. We could have used a step here where we change this to instead of the
137:17 entire user reply, we change it to an entity. Um, but it's not as reliable as
137:21 doing it this way. So, I'd prefer to just get it right the first time. Um,
137:26 cuz a entity of the number of uh number of feet may be a bit harder to pick up
137:30 than a clear word like an entity of house or apartment or a name, etc. So
137:35 just to make sure that it works every time for us, we want to create a new
137:39 prompt. And I'm going to put this in here. Extract the number of square feet
137:42 from what the user said in numerical format only. Each 500 include nothing
137:45 else in your response. This will be saved as a variable and passed to an API
137:49 to a quote generator function. So giving it a bit of context about what's been
137:52 going on. And also put in the memory there just for good measure. Oh, we didn't name the prompt.
138:05 And then we need to set the variable that we want to save this to, which is
138:08 going to be proper property size. So now we're saving that user's, the size of the user's property
138:19 in square feet. And then we need to do one kind of tricky step, but it's just something
138:25 that you guys will pick up as you go and uh as you build more of these. But the
138:28 next step after this is going to be sending that information to a relevance
138:31 tool. And the relevance tool is expecting not what's called a string,
138:36 which is just a number of letters or just text. Um, you can have a number as
138:39 a text, which is confusing, but basically it's just the format uh in
138:42 which it's being received in. So, because we're sending it over an API, we
138:46 need to be specific about the format. So I can't take this is essentially going
138:50 to be saving that 500 or say if we say it's a 500 foot property. This is going
138:55 to be plucking that out and giving us 500 as a string. In order to get the
138:58 response we want from relevance we need to convert it to a number and then send
139:03 it. So a little block of uh custom code here. Um know this was a no code
139:07 tutorial but I hope you can forgive me this. And all we need to do is put this
139:13 in here. So property size this is the variable. We're going to go property
139:19 size and we're just casting this variable as a number and then
139:23 reassigning it to the variable that it was before. So we're taking whatever
139:25 came out of this, we're saying, okay, can you just make it a number um and
139:29 we'll save it and sort of overwrite the existing variable. So now we have the number 500
139:34 in that case that we're ready to send to the next step. And then if that's gone
139:49 And then we have this JavaScript fail route here which you can just sort of
139:53 throw down there for now. Um there is a chance that this prompt outputs not just
139:57 a number. As you can see we're asking it for just this and therefore the property
140:04 size variable if it just is 500. Um but sometimes it can say hey your size is
140:11 500. and then we end up with a variable that's not actually convertible into a
140:15 number. Um, so we do need to add a little bit there as a as a potential
140:19 fallback. Um, probably won't be doing it in this video if I'm honest. It's fairly
140:22 basic. Um, but you would add this in here some sort of looping back in to
140:25 make sure that it is actually in the right format or just make sure that your
140:28 prompt is actually only you put like a strict instruction only output just the
140:32 number um so that you get less errors there. So the next major step after this
140:36 is to get our relevant AI research. If we go back to the Figma here, you'll see
140:40 that we have this relevant AI tool which is going to allow us to uh generate the
140:44 instant quote. So, I've pre-made this and I've actually sniped it from some of
140:47 my accelerator resources. So, if you open this link, it's going to allow you
140:51 to clone this into your relevance account. So, up here you can click clone
141:01 AI. And now this thing is going to be taking in a property type and a size and
141:04 square feet. And then we have a basic LLM step here. This is a really really
141:07 simple one. Again, like I said earlier in this video. The building really
141:12 powerful and advanced functionality does take a lot longer. And if I was to do
141:15 things that weren't just a very basic LLM step like this, then it would take a
141:18 lot longer with different platforms you have to sign up for. The idea is to
141:22 teach you how these things can connect. And then you can come into relevance
141:24 here. Once you know how to connect voice flow to relevance through an API call,
141:27 you can come in here and throw whatever the hell you want in and make it as
141:30 advanced as you want. But in this case, we literally could have done this in
141:34 voice flow if I'm being fully honest. But the idea of being able to access an
141:38 external tool via an API is really the skill that I'm trying to teach you here.
141:41 So, what this is doing is taking in the property type. We've got two options
141:46 here and the square footage in a number. See here, it's it's a number. I can't
141:49 just type in here. It has to be a number. So, that's what the API is going
141:51 to be expecting. You can see here that it says it's expecting a number. And
141:55 then we pass this into a basic prompt here that's saying the customer is
141:58 requesting this um and these are the rough prices for the different square
142:02 footage and different types. and the LLM in this case GPT4 mini just going to
142:05 look over that take in their inputs and then give an output. So if I just go up
142:10 here and I set it's an apartment and I mean I don't know what five how big is
142:15 an apartment in square feet I don't know that it's going to give us some kind of
142:24 output and then it gives us the estimate. So it's saying regular
142:26 maintenance cleaning for 60 deep cleaning for 120 move and move out etc.
142:30 So that's the output that we're going to be sending back to voice flow and we're
142:33 going to be turning that into a nice message to send through chat or through
142:36 the phone. So in order to set this tool up in voice flow and be able to interact
142:40 with it, we need to we can hide this make sure the tool's been saved and you
142:43 have it in your account and then you can go to use here and just like we learned
142:48 in theuh NA10 tutorial we can go to API here and then we get an API for us to
142:53 use and again is a post request and it tells us how to use it. So we have
142:56 params basically the inputs it's expecting of the property type and a
142:59 string. You see how it's got two uh quotations? That means anything inside
143:02 it is a string. And the square footage here, you can see it isn't in quotation.
143:05 So, it's not a string. It's in this case, it's a number. And of course, we
143:08 have the project ID here, letting relevance know which tool that we've
143:11 created that we're actually trying to interact with. And so, all we need to do
143:14 now to interact with this and set up our quote generator is to copy this link.
143:17 So, this is the endpoint URL that we're going to be calling over the internet.
143:21 And this is what it's expecting in the in the body of that post request. If you
143:25 go back to voice flow here and we need to go to dev API, we're going to change this to post.
143:31 Put in that URL. And in this case, seems that we've got a little bit of a different page on
143:36 relevant for some reason. So we can actually do the authentication step,
143:38 which is helpful. So you can click generate API key here. May have changed
143:41 by the time you're watching this video, but should roughly be the same. And we
143:46 can click deploy here. Make sure that the API is up and running for us to
143:48 interact with. Actually, now that we've deployed it, we don't need the authentication
143:55 anymore. That was a bit strange. Uh, actually, I will do the API key just so
143:58 you guys see how this works. We can make it private here. So, now that we have
144:02 our API key, we need to see how it's expecting to receive that API key via
144:05 the HTTP request, which we're going to be sending from voice flow. Um, so if we
144:09 scroll down here, curl is usually the one I like to go to. And there is a
144:12 little bit I haven't really explained in terms of headers and bodies when it
144:15 comes to API calls. But for now, just know that when you have a curl request
144:19 like this, maybe it's even easier on the uh JavaScript. Here we have what's
144:22 called headers. And this is basically like the the the envelope that you put
144:27 the information in. So this up here is the information that we're putting in
144:29 the request. It's going inside the envelope. And the headers and the method
144:34 and the endpoint URL are like the stamp and the information that you put on the
144:36 outside of the envelope to make sure it gets where it wants to go. So you can
144:40 say that this endpoint here, that's what we call the endpoint. Same as up here.
144:43 That's the the same as we have just here. Endpoint is like the address where
144:47 the envelope is being sent to. We have the post, the method. Maybe this is like
144:52 super fast mail or like overnight delivery or maybe like a parcel versus
144:56 an envelope. Basically, the type of delivery that you're doing or type of
144:59 request. And the headers include important stuff like the type of content
145:02 that's inside it. So you might say this like this is there's written text or
145:05 there's a letter in here. So it might seem a little bit complicated, but we'll
145:08 fill this all in. Now, anytime you are making an API call, you'll see these
145:11 headers around and you'll see the endpoint and the method and then the
145:14 body which we're going to be setting up. So, this will all make sense in 2
145:17 seconds, but let's just say for now we have content type and it's going to be
145:20 application JSON. That's one of our headers. So, if we go back to Connor's
145:25 cleaning, we open up the headers. We can go content type here as we saw here,
145:29 right? So content type, we need to set it to application/json. Very common one that
145:33 you're going to be using. And we have another one which is authorization. So this is basically the
145:39 majority of the endpoints you're going to be seeing which is authorization and
145:43 content type. And a lot of the time it's going to be an API key and application
145:47 JSON. So then we can go back to relevance. We can get our API
145:54 key. And now to set up the body of the request which is this. We can copy this
146:01 and change this to raw. And we'll paste this in. we have the params or the
146:05 inputs in this case, the property type and square footage for the tool. And
146:08 what we want to do is insert the variables that we've got over here into
146:12 these. So we can go inside these quotations here and we can go property
146:20 uh type in this case and then for square footage we can go property
146:27 size. And so because the property type is expecting a string as we can see if
146:32 we go back here um to build It's expecting a string and this is
146:37 expecting a number. The string must be wrapped around in uh these quotations
146:42 and the square footage is a number and so we can just put the number because
146:45 we've already converted it into a number here. Right? So that's probably the
146:48 trickiest part of all the stuff I'm going to teach you today. And now that
146:50 we have this set up, we can actually send a test request and we can say uh
146:58 house. We can say 500 and test this API call. And there we go. It's complete. We
147:02 have got back our answer the estimate etc. and all of the information is
147:06 coming back from relevant AI as expected and now what we need to do is extract
147:09 the information which we want which is this answer from this API call and from
147:13 the response. So we sent a request and we got a response. Remember how we were
147:17 using those terms before. We can click on answer and we can save it to and we
147:22 can call it uh raw quote data because it is quite raw from. And now we have the information
147:37 back from relevance AI. All we need to do is make it pretty and use AI to
147:41 generate a message that summarizes the quote to the customer. So we can go a uh
147:48 prompt step here. We can create a new prompt. Let's add in the conversation
147:53 history for context. And I've got this prompt here. Write a short and clear
147:58 explanation of this quote for the customers. uh we can go property property
148:06 type for the customer's property type and we can put in the raw quote data
148:10 here. So that's going to insert this with the customer's apartment or the
148:13 customer's house and then we're just going to dump in the raw data that came
148:16 back from relevance AI. Your response will be read over the phone. So it must
148:19 be all in one paragraph and no longer than three to four short sentences. It
148:22 should read like and I've given an example here of how I wanted to give the
148:32 And so we can change this to quote quote response. And now we are
148:38 ready to give this a spin. So if we run this whole thing from the top
148:50 run, how much for a weekly clean? So I'm going straight to the asking a question
148:53 about price. You're asking about a pricing for our weekly cleaning service.
148:57 Is that correct? Yes. Bam. Bam. Okay, sure. To give you an instant quote, I just need the property
149:03 type and size and square feet. Is the house. How many square feet is it? 500.
149:15 I don't know. Is that a big That sounds very small to me. I don't know if 500 I have I have no
149:20 I have no clue about the sizes of houses if 500 is normal or
149:26 not. And there we go. Based on your 500 foot house, we have four cleaning
149:29 packages. Regular maintenance cleaning at $90 for standard weekly cleaning,
149:33 deep cleaning, ra. So, we're giving them a quick summary based on their 500 ft
149:38 house or x number of square ft x property type and giving them a quick
149:42 summary. So, that's pretty cool. We're using relevance tool and we're getting
149:44 this information back. Now, there is one more step that I have added onto this
149:48 this I get this thing where if I've gone this far with you guys, I may as well
149:51 add in like the rest of it to make it actually a bit more useful. um which is
149:55 a quick lead capture using Google Sheets and Make.com. So, I couldn't leave you
149:59 guys hanging on this. I thought I may as well throw it in there. So, stick with
150:01 me because this is really where you're going to be like, "Oh, this is this is
150:04 uh opening my eyes to to what you can do with these kind of platforms." So, the
150:07 reason we're adding this on is because the person has asked about pricing or
150:10 they're directly interested in some kind of services and we've given them an
150:13 instant quote and now we're trying to immediately follow that up with, hey,
150:16 look, give us your details and we'll be in touch and we'll get that service
150:20 booked in right away. So, we can jump into a uh message block here. This is
150:24 I'll just paste this in to keep things nice and quick here. Please provide your
150:27 name and phone number and I'll get one of the team to call you to find a time
150:29 that works. Now, one thing I will change is as you saw on that last run, it had a
150:33 question at the bottom. It's like, which one are you interested in? I would
150:38 probably try to remove that. Um, do not ask a question at the end.
150:53 So that's just going to end the message and saying, "Look, this is a quote."
150:56 Bam. And then the next message I get is this. Please provide your name and phone
150:59 number and I'll get one of the team to call you to find a time that works. And
151:02 so we want to save this entire user reply with a capture step. Capture step.
151:08 There we go. Change this to entire user reply. So we want to capture all of the
151:11 information that they send or over the phone or either through chat. So in this
151:13 response, they're going to say their name and their phone number, right? and
151:17 we need to extract those out. I'm going to put this here and do a
151:23 set step. We're going to use a prompt. So, we're setting this with
151:28 AI. And we're going to create a new prompt. Um, let's just set the variable
151:32 first. Let's say this is going to be put here. And this is the prompt that we're
151:42 going to be using. So, we have just asked the user to provide their name and
151:44 phone number. We need to attempt to extract the information and then confirm
151:48 it with them. Here is their reply last utterance which they've just provided
151:51 and we've captured. If there is a valid name and phone number present, then you
151:54 must do a confirmation eg okay quickly to confirm your name is this and number
151:57 is this. Is that correct? However, if one or both are missing or appear to be
152:01 invalid, you must output only retry as your response and nothing else. This
152:05 retry variable will be checked and if it matches exactly, then it will trigger
152:09 another attempt to capture. So either a write a short and sweet confirmation
152:13 message or b output retry for another attempt at capturing. So what we're
152:17 doing here is using AI to analyze the response and say look we're looking to
152:20 pluck out a name and phone number and we want to also confirm that cuz this is
152:22 likely going to be over the phone. And if the AI doesn't see a clear valid
152:27 phone number and a valid name then it's going to output only the word retry. So
152:31 give us a response. If it's good to go and we can move on to the next step. If
152:35 it says retry then we're going to try to retry. Um, now of course this retry
152:38 doesn't actually do anything unless we build the functionality in to look for
152:43 that retry keyword, which we'll do in a second untitled prompt. We can change
152:54 name. And then we can go to um go if last response which is the
153:03 variable that we're saving the output of that prompt into here. Um whether it's
153:06 going to be retry or the say just to confirm is this your phone name and
153:12 phone number. Um if last response is retry or even just to make it a bit more
153:17 flexible contains retry unless unless the person's name is like Bill Retry
153:21 Smith then uh this should be fine. And actually, just to make this look a bit
153:24 cleaner, I might change this to um if last response um does not contain retry.
153:39 here and we are going to send the last response because as we said, it's going
153:42 to either generate the confirmation message or it'll output retry. So if
153:46 it's valid, it'll and we put up last response here. They'll say their name
153:50 and phone number. We'll analyze it and then we'll go great. It doesn't contain
153:53 the word retry in it. And bang. Hey, just to confirm this is your name and
153:56 this is your phone number. And then for the choices here, we can go and use our
154:00 handy dandy. They're going to be giving us a yes or no answer to this. So, we
154:07 no. And so, if they say no to the confirmation, say no, that's not my
154:11 correct phone number or name, then we need to have some kind of retry um logic
154:16 here. I usually like to make my retries um a orange color. And then my failure
154:27 say, "Okay, let's try that again. Can you please give me a full name and phone
154:30 number, please?" And then we're going to send them all the way back to this step
154:33 here. So they're basically going to recapture their information and then put
154:36 them through this process. And this is a loop that can be done over and over and
154:39 over again. So, and then we also need to deal with this else step. So if the word
154:43 does contain retry and it has said, "Hey, look, this isn't a valid phone
154:47 number or a name," then we need to deal with that as well. So, we can come down
154:52 here and go to message. Pop it under here. And I've got a message for this.
154:55 Sorry, I didn't quite get that. Can you please give me a full name and phone
154:58 number so a member of our team can get touch. Um, right click on this and you
155:04 can go block color. Change it to an orange. And then we're going to be going
155:08 back to here as well. So, it's helpful. You can click on these arrows. So, the
155:12 lines here, and you can change them to the same color. So, we want to make them
155:16 a bit more obvious that they're coming expected. We can make this an orange one
155:24 as well. And this one, too. So, if they come in and they say,
155:30 "Hey, my name's Bill and my phone number is 02111." And it comes in here and it
155:34 goes, "Hey, that doesn't look like it's proper." It's going to send a retry as
155:38 the output. We're going to pick it up here and it's going to say, "Sorry, I
155:41 didn't quite get that." and it's going to come back up and they're going to be
155:43 expected to give it again and it will go through and then once we got a valid
155:47 name and phone number and it's not outputting retry, then it's going to go
155:50 through here. It's going to spit that out and say, "Hey, just to double check,
155:53 this is your phone number and email before we proceed." Yes and no. No is
155:56 going to be handled there and it's going to take them back to the first step
156:00 again. So, that's some nice um error handling and sort of looping that you're
156:02 going to need to be building into a lot of your conversational AI agents,
156:05 especially on voice flow, right? And so, the last steps here are some quick
156:08 variable extractions. So we can go to div logic here and go to set and we're
156:13 going to extract the name and the phone number. So we'll go prompt. Holy moly,
156:19 it is it's bloody hot here. The variable that we want to save this, we were going
156:23 to be extracting the name. So we'll add name. Um, and this is going to be called
156:37 name. Here's a quick and easy prompt. You can pause it to take a look at that.
156:39 It's just going to extract their name. We need to add another variable.
156:53 prompt. Just paste this one in here. Pretty basic. Again, pause it if you
156:56 want to take a look. And I haven't named that prompt. It's going to annoy
157:05 me. Extract phone number. And now we're going to have their name and their phone number
157:11 extracted out of this response. And oh, and actually we need to add in
157:18 the conversation history there so that go great. Let me get that added into our
157:35 system. This buys us a bit of time as we use our um make web hook which we're
157:39 going to set up now. So the next step is to get a Google sheet set up and to use
157:44 make.com to uh take this data and shoot it into a Google sheet. So to do that um
157:48 I will leave a link on I mean you can just search it up. It's make it's
157:52 make.com. All right. I'll save you the hassle. So you sign into make.com create
157:57 an account whatever you want to do or Go to scenarios and then we're going to
158:08 create a new scenario. I'm going to build from scratch here. I'm going to get all that
158:13 rubbish out of the way. I don't know why it's acting like I'm some rookie here.
158:20 Um, and then we need to go to web hooks, custom web hook. We're going to
158:26 um we're going to add in a new web hook. This is going to be conors cleaning lead
158:32 capture. going to save that. It's going to create this web hook here. I'm going
158:37 to copy this edges to clipboard. We're going to come back to our build and then
158:41 we're going to go to the API step. So, what this is doing if
158:45 you're a bit new to to web hooks and and API calls and stuff. What we can do here
158:50 on make is set this up to basically listen. It's a URL. You know how we had
158:54 the endpoint? The end point, this thing here that it's given us that I've just
158:58 copied to our clipboard. That's like the address, remember? So if you put it
159:01 write it on the on the letter, that's where it's going to go. This allows us
159:05 to basically send data um via API call um from voice flow to make and it's
159:09 going to catch it here. And this little lightning bolt means that anything we
159:13 build after is going to be triggered whenever it receives one of those uh
159:17 whenever a new bit of mail arrives. It's going to then trigger this multi-step
159:21 process. So we're going to put this into voice flow. We're going to trigger it
159:24 and make sure it knows what data to expect. And then we're going to be able
159:27 to use that data and put it into Google Sheets. So stick with me here, but this
159:31 is another very very essential skill in AI automation is how to set up a webbook
159:35 um and use it within different apps. So we have our webbook here. We've copied the address to
159:41 clipboard. We're going back setting up a get request here. So it's not a post
159:45 request this time. We're just uh getting and we're technically not getting data.
159:48 A get request is a much more kind of quick and dirty request. Um and as
159:53 you'll see, we kind of just tack on a bunch of information after this. Um we
159:56 can do it through what's called parameters. Here we want to be sending a
160:02 property. Oh, let's just say property type. Actually, let's do this properly.
160:09 Let's go time um timestamp. So, in this Google sheet, you're going to want to know when the
160:16 different leads came in. So, we can go uh time stamp and get the time stamp
160:20 from voice flow. That's a default variable that they are automatically
160:23 filling out for you. So, one of the things, one of the rows in the
160:27 spreadsheet is going to be uh the time stamp. We're going to add another and
160:32 we're going to go um name. We're going to put in the name here. So, now all the
160:36 cogs in your head should start turning as we put this together. And we're going
160:44 to go uh phone number and we go bracket phone number number. And if we add another, we can go
160:52 property type. We can go size and go property size. We can add
161:12 another one. We can go quoted prices so that the sales team knows what we actually told them. in
161:18 case you're playing around with pricing. Um, raw quote data. And I'll probably
161:24 throw in one more here, which is their uh, user first reply. So, that might
161:34 give context to the sales team like what did they actually contact us for in the
161:37 first place? Maybe helpful, maybe not. But we can just send this all over to
161:47 And so now we can see this is uh as this thing's spinning around, it's basically
161:50 waiting for us to send some data to it. It's basically sitting there at the
161:54 mailbox like waiting for it to come through. Um and we can go send here and
162:00 I'm going to put in uh gosh dug myself a hole here. Um let's go name
162:09 Liam. Um house probably size. Um um lots of money. Um how much for
162:13 cleaning yarning? um send. And if we go back, bam, successfully determined. And what we've
162:20 done and determined means is that make has received the the the request that we
162:24 sent. And it now it knows that we're going to be sending it a time stamp and
162:29 a property type and this this really really key skill to understand because
162:33 now when I go oh save now when I go to here and I go Google Sheets and I go add
162:38 a row, um you will need to set up your uh Google Sheets connection here. So, you
162:44 just sign in with Google, add your connection in. Um, I am going to have to
163:00 quickly. I'm going go timestamp. So I'll zoom this up. Time stamp name
163:06 phone size. 1 2 3 4 5 6 7. Yeah, we've got them all. Right. So then I can
163:26 go. So call this my Connor cleaning support agent leads. So now I want to go back to make and I
163:37 want to click here to um Connor. There we go. Connor's clearing. Why do I keep spelling
163:49 cleaning? Now we have the spreadsheet set up. The sheet name is just going to
163:54 be sheet one. Sheet one. Does the table contain headers? Yes, it does. And now
163:58 we get to put in all of our values. So, bam. Time stamp. Pop that in there.
164:04 Name, pop that in there. Phone, pop that in there. Property type. So, you see
164:10 what such a key skill that uh I really, really want you guys to learn. Um,
164:14 because this is a lot of like if you just have a couple of these key things
164:17 using these APIs and say voice flow or in NAD, building tools and relevance and
164:22 then using them via API, knowing how to set up web hooks and then send data in
164:25 between them and to set this data structure. You send that first initial
164:29 batch over to make in a test test request. It's going to go okay this is
164:32 what they're going to send me in future and then that's locked in and you can
164:35 send thousands and thousands of requests through it and it will operate as you
164:38 expect and first question. So uh it's got all those other rows we don't need to worry about. Save.
164:44 All right. And so that should be working. We can switch this to
164:49 immediately. Oh, I need to save it. Save. Um I'm going to immediately as data arrives. So, this
164:55 is going to be waiting all the damn time. Actually, I'm I've got a whole
164:59 bunch of other stuff running in make for my my various businesses. So, I'm just
165:01 going to have this set up. If you wanted this to run around the clock, you turn
165:04 that on. I'm just going to turn it on to run once. And this is just going to be
165:07 sitting there waiting for data. So, what we can do is just finish this off
165:13 quickly. So, I'll just paste this in. Got this all done. Thank you. That's
165:19 the end of it. Um, what we want to do is have a failure path. So, in case this
165:32 say, "Darn it, something went red. Um, and then this one is like that.
165:42 All righty. So, if we look at make, it's still expecting us to send some data to
165:46 it. Let's hope that we've done all of this right and it's going to work first
165:50 time, but you never really know, dear. So, let's give this a run and we'll try
165:55 to go through that second path and we'll spreadsheet. I need weekend cleaning,
166:13 So asking a question is going straight to price. You're looking for weekend
166:16 cleaning services and want to know the pricing. Is that correct? Yes. So our
166:19 router is going to pick that up and say he's looking for pricing. Let's send him
166:22 to the quote tool. Okay. Sure. To give you an instant quote, I just need the
166:24 property type and size and square feet. It's a property. Let's go aart
166:30 mint. It's detected that correctly. And typical apartment size
166:47 feet. Okay, so 500 isn't actually ridiculous. Um, great. And then we get
166:50 our answer based on your 500 foot apartment. Bam, bam, bam. There's all
166:53 the prices. Please provide your name and phone number and I'll get one of the
166:55 teams to call you with a time that works. Let's uh give this a a test. See
167:00 if it's actually picking up on the fail. Sure. My name is and
167:14 is boom. Sorry, I didn't quite get that. So, it detected that it wasn't right.
167:18 So, we got the retry output. We got the retry output from this, which is what we
167:21 wanted. Sorry, I didn't quite get that. Can you give me again? Um Liam
167:36 021. That's what numbers look like here in New Zealand. Okay. Just to confirm quickly,
167:41 your name is Liam Mley and your phone number is that correct? Yes, sir. Name
167:50 number. Oh, what has it done there? Oh, dame. I don't know why it's I just want the
167:57 phone number. So, we have got a little bit of an error there. I just go back
168:00 and tweak the prompt. Make sure it's like only get the phone number. We don't
168:03 want anything apart from numbers here. Great. Let me get the added to the
168:06 system. And then if we go back to make, we see these are all green now. And we
168:09 see these dots. So, this is the information that came through. Namely,
168:12 Mly phone number. And so, here's the little mistake where we had a new line.
168:16 Apartment. Bam, bam, bam. All of that information. And then it's added it into
168:21 Google Sheets. Here we see updates, updated number of rows, all the values
168:32 to. Okay. Holy moly. Right. We're ready to put this thing on our website and to
168:36 also put it on a phone number. So, let's just finish the job. Guys, I'm uh
168:40 getting real hungry, but we can uh we can push through. So, I will just turn
168:44 this on actually so that if we are testing it on the web and over the
168:49 phone, um it's ready to receive. Um, if we do want to be pedantic, I would go
168:54 back and I would change I have to do it. It's going to piss me
168:59 off. I'll put their name. Oh, that's why. I'll put their phone number only.
169:14 only. And we do have these fail points here. Um, I'm not going to bother
169:17 filling them out. I think you guys can figure out based off how I've handled
169:21 this, how you can handle these as well. So, what you'll find is when you're
169:24 building these, these kind of fail like error handling um is kind of a an
169:28 endless thread that you keep pulling. It's like, oh, well, now I've got to
169:31 handle this, this, and this this. So, um I'm not going to this is a prototype.
169:34 I'm not going to be doing all of the the error handling for you here. In the
169:37 template, there is actually a little bit more of it. Um some better examples. So,
169:41 maybe if you import that, you can just steal the work that I've done there. But
169:45 what we need to do now is we can publish this thing. We'll call it
170:03 drop. All right. So now it's published. We can add the agent to a website. Let's
170:06 click on that. And that takes us to this integrations tab. Um let's put this down. I don't need to
170:13 see that. Um, they've got a new version of it. That's good to know. I said
170:16 installation is pretty straightforward. So, we can just click copy here. And
170:24 up I'm going to open up brackets here just to give you a demo of a website. I
170:27 use this in all my tutorials. It's really easy to spin up. Um, I will leave
170:31 a link to this template if you want it. Um, and also some instructions on how
170:34 you can open up a website. I know this looks like code and it's all scary, but
170:38 um, this is just allowing me to spin up a website very quickly. So, I'll leave
170:42 the template file. All you need to do is once you've downloaded the template
170:46 file, you need to download brackets, which is the software. You can go file,
170:50 then open folder, and then you want to click on the folder when you've unzipped
170:53 it, and it's going to open up the whole folder. And then you'll get all of this
170:57 uh opened up like this. And you see all of these files ra. All you need to do is
171:00 click on the index.html. And then you'll see something similar to this. Well, I'm
171:05 going to scroll down to the bottom of the index.html. I'm going to delete this
171:10 old voice agent I was testing on here. Drop this in here. Paste that. And then
171:14 save it. Command S. Click this little button up here. And it will show us a
171:20 local version of the website running on our computer here through brackets. So
171:24 here's my man with a magnificent beard. And we have the tester agent bubble down
171:40 I want to know where you Yep. Woo. Okay. Uh I didn't even program
171:54 that in there. Maybe you just thought it was inappropriate. Um but we've got it
171:58 working on a website. Now, if we pop back over to um uh Voice Flow here and
172:04 you go to the integrations, the widget, you see we've got this test your agent
172:07 thing. So, down here, we can play around with the look and feel of it. I'm not
172:10 really going to get into this here. There's quite a lot to play around with,
172:14 but basically all of what you see on here can be changed around. Different
172:18 logos, different text here, different icon, etc., different colors, and you
172:22 can just make it look and feel however you want it to. So, I'm sure you guys
172:25 are big enough and ugly enough to figure that out yourself. we'd probably want to
172:29 switch over to uh this here. One thing you would want to do is turn off powered
172:33 by voice flow so it's not uh sending traffic to them when it's on your own
172:36 website. And that's about it. For the sake of time, I'm not going to go
172:40 through the entire flow again here. Just know that the functionality that we
172:42 built that I just showed you in the builder is going to work cuz we just
172:45 deployed it. Like this is exactly what we're interacting with. So, it's all
172:49 working here. The only step to do now is to put it on this phone number so we can
172:52 have a chat to it over the phone, which we're going to do now. To do that, we
172:55 need to go to the telefan bit here. It is in beta right now, but for most of
172:58 you watching it is not going to be by the time you you are watching this. So,
173:01 we need to set up a phone number from Twilio, import it, and then connect our
173:04 agent and its functionality to that. So, we can go import number. You'll see that
173:08 we have this information here. So, we can use Twilio or Vonnage. Twilio is
173:12 usually the go-to here. So, if we click on learn more, then they're going to
173:15 help us. Basically, the best way to make sure you're getting the most up-to-date
173:18 information is go to the docs of the platform. Finding and reading and
173:20 extracting information from documentation on these kinds of platforms is another key skill that you
173:25 need to pick up to succeed in the space. So if we go to the docs here, we go to
173:34 um voice phone number setting up Twilio integration and they have a video here
173:37 adding a phone number to your agent. So if you ever get stuck, you know, you've
173:40 got documentation here and for all of the other platforms, but they'll keep
173:43 updating these videos if things change, which they likely will as this voice AI
173:48 space really takes off. So, if we go to Twilio, you will need to sign up and
174:06 Twilio. All right, so we are logged into Twilio. You'll need to create an account
174:10 for most of you, but Twilio is a uh phone number provider that you can
174:13 connect to and interact with over the internet. super helpful when you can buy
174:16 lots of numbers from different locations and stuff. When it comes to phone
174:19 numbers, it can there's a lot of rules and regulations around different like it
174:23 varies a lot from country to country. So depending like if you're in Germany, I
174:26 believe in order to get a German phone number, you need to have a company
174:29 registered and get the number through your company registration and provide
174:33 those details. So can be difficult. I'm just going to show you how to use a uh a
174:36 US-based number here. So we can go over to the phone numbers on the left here.
174:40 There may be some setup that Twilio walk you through. It can be kind of annoying
174:43 sometimes. They say you need to do all of these declarations and forms and
174:46 stuff, but for the most part, it should be fairly straightforward if you follow
174:49 their setup instructions when you create your account to then come over and go to
174:54 your phone numbers and manage and go to buy a number. Now, unless you have other
174:59 purposes you want to use this for, you can just snag any random one if you're
175:02 following this tutorial. Um, if you're obviously doing this for a client, you
175:04 could get one that's matched to their location or their their state or even
175:08 their city. And when you click buy, you can see there's all these kind of
175:11 registrations and RAR you need to do. But thankfully, voice is uh is ones that
175:15 don't need all of that. And you've got global routing, etc. So, you can come
175:18 down here and buy this. It's going to be a dollar a month. I know cost of
175:21 starting up a business is ridiculous these days. How dare they? But just to walk
175:26 things through and do it with you, I'm going to buy this number even though I
175:29 really don't need it. There we go. I've got my number. I'm just going to copy this so I don't
175:39 forget it. 517. And then I go to active numbers. I got my guy. Where is he? 517
175:44 down here. Copy this bad boy. Go back to Twilio. Put our number in here. Our
175:51 Twilio account SID. You need to pop back to Twilio. You go to your account
175:56 dashboard. This was a bit confusing when I first did it. So, I'm glad I'm glad I
175:59 can help you guys through it because it was a bit of a pain in the bum for me.
176:02 You think all this information is going to be like tucked away in the side
176:05 panels, but no, it's just on the dashboard. Who the hell puts your or
176:09 tokens and stuff on a dashboard? I mean, it's when you know it's there, it's
176:11 helpful, but it was a bit confusing at the time for me. Um, I'm going to paste
176:22 here. We're going to copy this or token and paste this in here. And we're going
176:29 to say this is my you YouTube tutorial. And I'm going to import that bad boy.
176:33 Select environment. I'm just going to put it on development for now. uh we can
176:36 assign that. And with that, you can see if I click unassign here, that means I'm
176:41 going to take it off. So, it's it's assigned and it should be ready to have
176:44 a chat with. So, if I click to copy this number here, I go back to my little
176:52 uh uh website code here. If I go find um plus one. Oh, there we go. So, we have
176:58 the call. Um this is in the header. Uh this is just me p wanting to put it on
177:01 the site for you guys. So, you guys don't necessarily need to copy this. I'm
177:05 going to save that. If I go back to my website here, now I have the actual
177:10 phone number. And if I pull up Skype on my phone, you guys, if you're in the US,
177:13 you can just or wherever your number is based, you can just call it stage plus
177:20 one. All right, let's give our mate a Welcome to Conor's Cleaning. How can I
177:28 help? Hey, uh, I was just wondering, uh, business. Is that correct? Yeah, that's
177:52 We're located at 247 Beacon Street in Boston, Massachusetts 0216. We serve the
177:57 greater Boston area, including neighborhoods like downtown Boston,
178:01 Cambridge, Somerville, Brooklyn, and several others. Is there anything else I
178:05 can help you with today? Sweet. Okay, so that's done the uh knowledge the knowledgebased route.
178:12 Now, let's give it another spin. Call Hey, welcome to Connor's Cleaning. How
178:20 can I help? Hi there. I'm just getting interested in getting uh some cleaning
178:30 It sounds like you're interested in booking a cleaning service for a
178:43 Okay. Sure. To give you an instant quote. I just need the property type and
178:47 size and square feet. Is the property a Sorry, I didn't get that. Is it a house
179:08 And how many square feet is it? It's One sec while I generate your quote.
179:20 Based on your requirements, we can offer four different packages. Regular
179:24 maintenance cleaning at $90, which covers standard weekly or bi-weekly
179:30 cleaning. Deep cleaning at $150 for a more thorough top to bottom refresh,
179:36 move in, move out. Cleaning at $170 for preparing or cleaning after a home
179:40 transition, and post renovation cleaning at $200 for handling construction dust
179:45 and debris. Please provide your name and phone number and I'll get one of the
179:48 team to call you to find a time that works. Sure. My name is Liam Otsley and
180:05 Okay, just to quickly confirm, your name is Liam Oxley and your phone number is
180:11 02119792. Is that correct? Yep, that's Great. Let me get that added into our
180:24 system. All done. Thank you. Boom. All righty. That is done. That is dunzos.
180:32 All right. So, you guys learned a lot in that one. voice integration, website
180:37 integration, um connecting web hooks to make, setting up make automations,
180:41 sending data over, connecting relevance AI tools into into voice flow. Um basic
180:47 integration with a CRM, in this case, it's a sheet, but there's so much in
180:51 there, guys. I hope you really really uh learned a lot from that. This has been a
180:54 big one. And we've still got uh one more to go. So, I hope you're sticking with
181:00 us. Um but going back to our Figma here, um we have ticked off all of this. So,
181:03 we have it as a web chat widget and we have it as a a phone number. Now, as far
181:07 as I know, you can have both options for the same agent on voice. You can have it
181:10 on the website and over voice. You don't need to duplicate it and sort of define
181:14 what modality it's going to be. So, we've ticked off all the boxes for this.
181:17 All of the resources will be in here. All of the prompts, a template for that
181:20 whole final build as well. If you just want to snag all my hard work and go and
181:23 sell it to someone, again, I don't I really don't care. Um, that's what these
181:26 videos are for. And we're getting into All righty. So, last but not least is an
181:38 agent built on my own software. So, I didn't want to make this. It's not about
181:41 me selling you or getting you to use my software. So, I thought I'd put it at
181:43 the end just so you know that I wasn't really This is about you guys learning
181:47 and my software happens to help you put an agent onto WhatsApp very very easily.
181:50 So, that's why it's included in here. But again, this is nonsponsored,
181:54 nonpromoted, non whatever. I'm just really trying to share with you what I
181:57 think is a really valuable skill set to have. All right. Now getting into AI
182:00 agent build number four. This is going to be tada a WhatsApp based ARI customer
182:06 support and lead generation agent built on agentive my software. So this is a
182:11 noode uh AI agent builder that is built on top of OpenAI's assistance API. So
182:15 you're technically using your OpenAI account getting very very cheap rates on
182:19 the uh token usage that you're running through this agent. But Agent just
182:22 allows you to build on top of it very easily. but more importantly to deploy
182:26 these agents not just onto web chat widgets like we've done with voice flow
182:30 but easily onto things like WhatsApp and Instagram etc. So that's really the key
182:33 thing that Agenda focuses on doing right now is making it easy for you to get
182:36 your agents onto these platforms. So as you can see it's a fairly similar build
182:39 to what we just did on voice flow in terms of functionality. We're going to
182:43 be having a uh a knowledge base that we can ask questions over. It's going to be
182:46 able to generate another instant quote. So we'll just quickly connect that same
182:50 relevance tool here. And finally, we're going to do a lead capture, but this
182:53 time it's going to be done through Air Table. So, I want to mix it up and show
182:55 you how you can connect your agents to Air Table, which is a very, very common
182:58 integration that you're going to need to know. And the difference between this
183:01 agent, you're going to see that it's much much faster to build. This is not
183:05 meant to be a side-by-side comparison of what's better, how much faster. It's
183:08 just that when you build on a more conversationalbased uh AI agent platform
183:12 like agentive which is built on top of the assistance API, it's a very
183:15 different way of building agents because it's all just based on a prompt and
183:18 providing the right tools and all the magic kind of happens itself through the
183:22 prompt. Whereas voice flow gives you a lot more control. So it's really
183:25 difference between structured AI agent building versus more conversational and
183:30 open-ended chats through more chat GBT like experience that can just go on and
183:33 on and on which is what these agents can do. So the purpose is of course fairly
183:36 similar but the value of this is slightly different in that we are using
183:40 WhatsApp. So uh many people browsing for services online are hesitant to use
183:44 website contact forms or other or chat bots that they think are not going to
183:47 give them access to a real human due to the potential delays that come from it.
183:50 Right? You land in a website and you're you're shopping around for a different
183:53 service or product and then there's this this contact form or there's a a web
183:56 chat widget and you're going to go well I don't really think I'm going to get
183:59 the help that I really need here at the at the speed that I want. So you might
184:02 look for a WhatsApp widget and you know that if you click that WhatsApp widget
184:05 you're going to get to speak directly to someone and this is kind of playing on
184:08 that fact that if you have the WhatsApp option on your uh website people are
184:11 much more likely to just click that and go through and try to have a
184:14 conversation directly to get what they want. So by having a WhatsApp option on
184:18 a website or other triggers eg you can have a QR code that you could stick on
184:21 say a real estate sign and you build an agent connect it to your WhatsApp number
184:25 like we're going to do here and then you create a QR code that people can scan
184:28 and immediately open WhatsApp and start chatting with it. There's lots of
184:31 different ways that you can have an access point into a WhatsApp agent like
184:34 this. But it basically opens up more conversations through a more smartphone
184:38 native platform. So they can hop on their phone and sort of have a chat away
184:40 to it rather than being on a website on the computer or a little tiny website
184:45 chatbot on their phone. Um, in order to essentially engage more prospects or
184:48 more people interested in the business in conversation, quickly provide value
184:52 through either the knowledge base and these tools here and real-time quotes.
184:56 And ultimately, because you're providing that instant value and instant feedback
184:59 from them, collect their lead information, or better yet, even set
185:02 appointments through WhatsApp, which you can build on agenda. But that use case
185:05 is a little bit more advanced and not something I can show within this video,
185:08 but it definitely is possible. But it's really only a few steps away from the
185:11 skills that you've learned in this video so far. So keep an eye on that
185:14 appointment setting use case because if you can do that with AI agent, it's a
185:17 very, very valuable one. And I've done other videos on the channel here showing
185:21 you how to do that. So, the usage of this is that they're going to find the
185:24 uh company's WhatsApp number on their website or elsewhere, maybe a QR code
185:27 like I said, and they're going to start a conversation on WhatsApp, and then the
185:30 agent is immediately going to jump in and start responding and be able to
185:33 answer from the knowledge base, generate quotes, and then capture their lead
185:36 information. So, without further ado, let's jump into building this agent. So,
185:39 we can click up here to go to my website, Agentive. You can click on
185:42 register now. You can just register with your Google account. I'm going to log in
185:50 It is free to make an account and we have a free plan so you can just
185:52 experiment around as much as you need and then you're only going to be charged
185:55 based on the amount of usage you use. So it's very very cheap and affordable and
185:58 I wanted to make this platform for you guys to all get on and experiment with
186:02 building AI agents without coding. That was really the the core of why we
186:05 started this whole thing. So we've got the dashboard here which will load in my
186:07 data in a second. So you can see here what the dashboard will look like when you've got
186:12 your own agents running. We are running the Agentive customer support chatbot
186:16 through this uh through this account that I'm I'm using right now. So you can
186:20 see usage costs very cheap sessions etc. So it's really cool when you go into
186:23 analytics and you can use agentive to see how people are using your agents but
186:26 that's obviously something for a little bit later once you put these into
186:29 production. Uh now what we're going to do is of course we can go to agents or
186:33 we can just create an agent from here and I call this um Connor's cleaning
186:39 WhatsApp agent. Oh we got a little description as well. um answers
186:50 time so answers questions from the knowledge base provides real-time
186:54 cleaning quotes and can capture leads to air table. So you're going to see the
186:57 setup is a lot faster than some of these other platforms. Again, I'm not not
187:01 trying to gas myself up here. It's just a different way of approaching u
187:04 building agents. So it's a lot more fast and and rapid prototyping and easy to
187:08 get things up and running. Of course, if you need much more advanced
187:10 functionality, you do need to go the extra mile and go on to platforms like
187:13 Voice Flow. But in this case, we have a prompt, very easy. We have a knowledge
187:16 and we have tools. So, remember when we went back to being a a chef and the
187:20 three ingredients concept, these are your three ingredients, right? The
187:23 prompt that you get to provide as the instructions, the knowledge that you
187:26 provide as the external knowledge base and the tools that we can connect to it
187:29 as well. And we can select the model here. So, I think I want a nice and
187:33 snappy response time because this is going to be on WhatsApp and customerf
187:38 facing. So I'll go to GPT4 mini. So it's nice and quick. Now we can just put a
187:46 help. Just put that in there for now as the prompt. The knowledge base we can
187:50 turn this on. We can create a new this. We're going to click here and
187:59 we're going to upload that same file that we used on voice flow. The same
188:01 document that will be available in the resources for this uh for this
188:05 particular guide. Give that a second to process. Once this goes green, we're
188:07 good to upload it. And you can add multiple files in here. We allowed five
188:11 files at a time, but you can add dozens and dozens of files. So, you have a
188:17 with. And just like that, we have connected our knowledge base. And the
188:21 cool thing about Agent is because we're built on the assistance API, this is
188:24 actually an independent knowledge base. So, you can create a knowledge base and
188:27 connect it to multiple different agents. The the knowledge is not restricted to
188:30 the agent that you build it within. So, I can go and create a new agent and
188:32 connect this exact same knowledge base. And I have all of these other ones here.
188:35 And then when we go into the tools section, we are going to have two tools
188:38 for this. Well, the knowledge base, if we go back to our Figma here,
188:41 technically the knowledge base is a form of tool that the agent is using. But on
188:45 platforms like the assistance API and and many platforms, you'll see knowledge
188:49 treated as its own thing, but essentially it is just another tool that
188:52 the agent is using at the right time when it needs to pull in knowledge to
188:56 answer questions. So, OpenAI separates it out into its own thing here. And so
189:00 do we because we built on top of it. So we do have three different tools but
189:03 knowledge is its own tool that gets set up through this knowledgebased um
189:07 connection that we just made before. Then we have the tools and here we have
189:11 our instant quote from relevance and we have our capture lead information. So we
189:14 know the process of going on to relevance. So we can just go create a
189:17 new tool here and this is going to show you a schema. Remember back to when we
189:21 talked about schemas it explains to the agent how to use the tool. So to add a
189:25 tool to this agent we need to add a schema to it. And thankfully our buddies
189:29 at relevance AI provide a very very easy way to create schemas to import into
189:33 agents like on aentive. So here I can grab that same cost estimate tool for
189:36 the instant quotation for cleaning services that we've used previously.
189:39 Again this will be linked. You can just clone this if you haven't got it
189:42 already. I will provide a link for you to clone this into your relevant account
189:45 that will be in the resources for this video. And it's just like the previous
189:48 tutorial that we did where it's got property type square footage and an LLM
189:52 step here to calculate it. It's going to spit that back and we're going to turn
189:56 that into a nice response uh with our agent on Agent. We can make sure that
189:58 we've saved this. So, the cool thing here is that in order to get this
190:02 connected to uh Agent and our agent over there, we can just go to custom actions
190:06 here on the tools page. As you can see here, it's mainly intended for use with
190:10 OpenAI's custom GPTs, which you can get access through chat GPT. And I highly
190:14 recommend you do check out the OpenAI GPTs because it's a super simple way to
190:17 spin up your own agents um on the chat GPT site. And so, we can select our tool
190:21 here and we can get a schema for it. But what I've just realized is that I
190:24 actually do have an air table lead capture tool here that I've already
190:26 created on relevance. And it's actually going to be easier for us to set it up
190:30 here on relevance than to have to do it all separately. So let's just quickly
190:35 set that up. Now if we go air table, let's just get a simple one that
190:38 captures the name and phone. I'll provide the template for this tool so
190:41 you can clone it in. But basically it takes an input of the name of the lead
190:44 and the email address of the lead and the phone as well. So, it's capturing
190:48 all three of these as lead information and then it's sending it over to Air
190:51 Table which we're going to set up just now and it's using a post request to
190:55 push that data that we collected here and we will collect through WhatsApp
190:58 eventually and it's pushing it into the Air Table database. So, let's get that
191:07 Table and I'm just going to use a dummy CRM that I use for all of these
191:10 tutorials and you guys are going to be able to clone this if you want. In the
191:12 resources for this video, there will be a link like this. So, if I share this um
191:20 publicly, you guys will get something that looks a bit like this and this
191:24 button up here says copy base. That will copy it into your account. So, all you
191:27 need to do to copy this air table base is to create an air table account and
191:30 then click on this copy base and it will copy it over. So, you can get this with
191:34 the column source preset up. It is fairly easy to set up these fields
191:36 yourself, but I want to make it easier for you guys. So, you can just copy
191:38 this. It'll be included in the resources. But here we have the fields
191:42 that we're looking for. So now all we need to do to send data into this
191:45 database through our WhatsApp based agentive agent. So when someone provides
191:48 their details that it gets shot into here is we need to go and see our
191:52 details for the air tableable web API. So air tableable has their own API which
191:56 allows us to interact with our databases like this programmatically. So all we
191:59 need to do is go up to the right hand corner here go to builder hub and if we
192:04 go to the developer docs here and scroll down to the web API this is a reference for the air
192:10 table web API. So this documentation is essentially going to tell us how to
192:13 interact with our air table programmatically through our agents and
192:16 through voice flow and through relevance and and through agentive as well. So any
192:19 way you want to interact with it, you can now take this knowledge that you've
192:22 gained in this video look through this and model what we're going to do here in
192:25 relevance. You can take that same idea and put it into say voice flow and you
192:28 can build an air table integration within voice flow yourself where you can
192:32 send and pull data. So these skills all stack on top of each other and it really
192:35 centers around understanding how APIs work and that comes down to reading
192:39 documentation as well. So in this case, if you go back to our relevance tool
192:44 here, we need to get our URL, which is our endpoint, which we've talked about
192:46 before. This is the address that we are sending the request to and agentive
192:50 where we're building the agent. It's going to be using relevance to call air
192:53 table. It's a bit of a a roundabout way of doing things. But to get all of this
192:57 information, the easiest way is to go back to this documentation. And the
193:00 easiest way for us to find the information that allows us to interact
193:03 with our own Air Table base that we're setting up, is to come down here and
193:07 find the base that you've just cloned into your account, which will likely be
193:10 Smith Solar CRM. Don't worry about the name. And Air Table does a really,
193:13 really good job of making this super easy. And that we can just come here to
193:17 the leads table. So, if we go back to um Air Table here, you see we are on the
193:21 leads table. We have these different tabs. You can just ignore these are just
193:24 different projects that I've done on YouTube. Um they're all kind of there in
193:27 case people are also cloning this into their account. But we're looking at the
193:30 leads tab here. So, we go to the leads here and we want to create records. And
193:34 then it gives us all of the information we need here in order to create records.
193:38 So, um you can see that it's a post HTTPS and all of this information. So,
193:45 this is the endpoint. We want to copy all of this all the way down to leads.
193:50 Copy this. Go back to relevance and paste this in. Oh, maybe
193:55 that was already there. And paste that in there. And then we need to add in two
193:59 headers. So we have authorization and content type. Now remembering what we
194:02 learned before, you can see we have H and this H tag means that there's a
194:06 header. And so the header is authorization. And then the value is
194:10 going to be bearer and then our token. This is something that tripped me up
194:13 when I was first learning this uh using APIs. Is that you need to add this
194:18 bearer word and then a space and then your API key. It's a weird way of doing
194:21 things. I don't really know why uh why it's like that but sometimes when you're
194:25 doing these authorizations you need to add in bearer space and then your API
194:29 key. So it's it's there for a reason um is what I'm saying. And then we have the
194:33 content type being application JSON. So we're already familiar with that. So
194:36 going back to relevance we have the header of authorization and content type
194:40 here. Application JSON. And now we need to add in our Air Table API key so that
194:44 we are authenticated and we have permission to send an API request. So
194:47 they're not going to let anyone use this details and and start sending data to
194:51 our our database, right? They need to be authenticated and that's what API keys
194:55 do. So to get our Air Table API key, of course, we go to Air Table. We can come
194:59 up to the top right here, go back to our builder hub, go to personal access
195:06 tokens, and we can create a new token. Call this YouTube. We can go um add the base. This
195:15 will be um Smith's Solar CRM. We can add a scope read write and sometimes I find it handy
195:22 to have the schemas read in there as well. So basically what we're doing here
195:25 is saying that I give this API key that we're creating permission to interact
195:29 with this uh this air table and I give it permissions to do these things like
195:33 read what's in the database write to the databases and create new things and also
195:36 to see the overall structure of the base and the field types. So we can add that
195:41 and create the token. We get this token, head back to relevance. And again, this
195:45 template so that you can clone it into your account is going to be on the uh on
195:48 the resources. So, if you're following along, you should just clone it into
195:51 your account and then come down here and make the changes as I do them. So, we
195:54 can add a make sure we have a space after bearer and then paste our key
195:57 because if we go back to the web API docs, we can see we have authorization
196:02 bearer space your API key content type and then application/json. Then we have
196:05 the data as the payload. Remember, like this is what's inside the envelope. Then
196:10 we have records and we have the fields name, phone, and status. And then it's
196:13 provided us an example of how we would send data into that which we don't need
196:16 to worry too much about because I've already got this fitted in here. It can
196:19 be quite fiddly. In fact, for this I'm actually going to add another field in
196:23 here. This is one thing about relevance I'm not a huge fan of. This can feel
196:27 super fiddly sometimes. So if we go email, this will already be in the
196:34 template that I give you, by the way. Okay, so we have the URL has been
196:47 updated. The method is post. That's correct. We have the authentication. We
196:51 have the content type. We have the body all set up. We've added in our fields of
196:56 name, phone, and email, which we have name, email, and phone. So, we can give
197:01 it a spin here. If we say and we give it a spin. Run the us. Yep. And there we go. If we go back
197:22 to Air Table, open the base up. There we have it. Liam phone email.
197:29 So, we can take this tool and we can take the instant quote generator. we can
197:33 put those into uh Agent and before you know it, we're going to have our agent
197:37 ready to go. So, let's head back to relevance here. We'll save this tool and
197:41 we'll change this to name, email, and phone. Um, and just quickly
197:46 before we do that integration, this is when the description comes into play
197:49 here. Remember those natural language descriptions of what the tool does, what
197:53 each of the parameters and inputs are. It's really important to get these right
197:56 in relevance, and I see a lot of people skipping over this step, but this is
197:59 what's going to be put into that schema, right? So when relevance generates a
198:03 schema for us, that onepage manual on how to use this tool and use the API in
198:07 order to interact with this functionality when we give it into
198:11 agentive, it's going to be reading over everything in there. And it's going to
198:13 be those little descriptions around what the tool does and what it's supposed to
198:17 take in. And and these parts here in relevance is where we get to set that
198:20 up. A proper description is needed before we do this integration. So this
198:23 tool captures lead information, stores in Air Table CRM, requires lead's name,
198:27 phone, and email. Name, phone, and email. The name is name of the lead.
198:31 Yep. Email, email address of the lead and phone is the phone number of the
198:35 lead. So that's all good there and ready to integrate. Might even do a quick check
198:47 well. Yep. Type of property, square footage of the property, and we are
198:51 ready to go. So now we can click on the custom action step here. Scroll down and
198:59 click on both of these. Bam. Bam. Scroll down. We're going to change this
199:03 to custom orth. We're going to generate an API key. There we go. And we're going to
199:10 generate our open API, not open AI, open API. It's essentially a type of API and
199:14 a way of describing how the API works. And it gives us all of this information
199:17 here. I will actually expand it out so you guys can see at least some of it.
199:21 It's probably easier over on agentive actually. And then we head back to
199:25 agentive. What we can do is paste in the schema. And now if we scroll through
199:31 this quickly, I just want you to see what a schema looks like under the hood
199:34 because we have some important parts uh that's going to really connect the dots
199:37 for you after everything that we've learned in this video. So I'll zoom in a
199:41 bit here. Um we have the title of the tool. So we have a few key things in
199:44 here that we can break down. Basically the paths. We have two paths in here.
199:48 This is one of them and this is the other. These represent the two tools
199:52 that we are integrating. You can see one here is the operation ID is basically
199:56 the name of the tool and that is taken from relevance directly the air table
199:59 lead capture and the summary here is the actual name or the title of the tool
200:02 that we had in relevance. This is just a a version where they put in um
200:06 underscores to connect the uh the gaps and the description here you can see
200:09 it's the same as a description that we set up over I don't want to go back on
200:13 there but that was a description that we put under the name to describe what the
200:16 tool does and then as for the inputs relevance has made it a little bit more
200:19 complicated by putting a schema in here. Um, so we'll cover that in a second, but
200:22 basically here's the second tool. Sparkly cost estimate. This tool does
200:26 this. This about estimating the cost of an apartment. Then down here we have the
200:31 schemas for the inputs. So we have things like the name. This is for the
200:35 lead capture tool. The name um this is one of the fields. It's going to be in
200:38 type string and it's required. We have the email which is type string which is
200:42 required. And we have the email which is a description here. And of course you can see all the
200:47 descriptions that we put in on relevance showing up here. then the phone number
200:50 of the lead, the email of the lead, etc. And here it's specifying how the AI
200:54 agent should be sending inputs into that. So that's probably the most
200:56 difficult technical part of this whole video, but I did want to give you a bit
200:59 of context on how that kind of fits together. This is quite a complex
201:03 schema. Relevance puts it together in a little bit more complex way um by using
201:07 these uh these schemas for the inputs down here. But long story short, if we
201:11 then go to the add or button, we need to set up our authentication, which we can
201:15 do by coming back to relevance and copying this. We go back to here, paste
201:22 this in. We go custom orth and we go authorize a orization with a
201:30 zed. Oh, I need to create the tool. Sorry. So, we can just click create
201:34 tool. So, the tool has been created successfully. And there we go. We have
201:38 both of the tools added in because we did them both in one bundle on
201:42 relevance. And then if we go edit off, we can then put in API keys for both of
202:05 one. And there we go. Now we have our knowledge set up, our two tools set up,
202:09 and you can see that we're pretty darn close to completing this build. We have
202:13 all of these three done. Might as well make them green for the sake of it. And
202:17 now the only thing left to do is to write a prompt that connects this all
202:20 together. And that's really the glue that holds it together. My go-to method
202:23 of rapidly creating prompts for AI agents is using a relevance tool. Um,
202:30 perfect that I've created, and I I said I'd give this to you guys for free as
202:33 well. That's going to be included in the resources. But if I go to use
202:38 here, it's a prompt writer that includes all of the information from how we do
202:41 prompting at Morningside, which is based on research and includes all the key
202:45 things like RO, task, specifics, context, um, explaining how to use the
202:48 tools that it's been provided as well. So, I'm just going to fill this out
202:50 quickly here and then get a prompt. And you guys can steal my prompt from the
202:54 resources or you can use this as well to create your own. But, it's a pretty good
202:57 exercise because you can see here we Um, and what I like to do here is, so
203:07 this is just a quick rundown of what the agent does, where it is being deployed,
203:20 why. So you can pause the video and look at that there. But just a bit of context
203:23 on what the agent does, where it is being deployed, and why. Conversions
203:45 contain. Then we get to the tools say and then we have the other tool
204:09 And then for the ideal input and output examples, I'm just going to say none to
204:20 assistant. And so just like that, in maybe a few minutes, I've typed in all
204:23 of this information about the agent and what it does. And now I can just click
204:27 run tool here. And it's going to take all of this information, run it through
204:31 the prompt that I've written that bakes in the best prompting practices for AI
204:34 agents from my experience and from the projects that we do at Morningside AI
204:37 and also all the research that we've used to make those prompting practices.
204:41 And it's going to spit us out an AI agent prompt that we can throw straight
204:44 into Agent and it'll just glue everything that we've done together,
204:47 tell it who it is and what it's trying to do, tell it how it's supposed to use
204:50 the knowledge base, and tell it how and when it's supposed to use those tools in
204:52 order to reach its objective of capturing those leads for us via
204:56 WhatsApp. And there we go. So, if we scroll down, we can see it's spit out
204:59 this entire prompt. I'm going to change it to the raw text so we get all this
205:04 markdown formatting included. We can view all here. I'm going to copy it all
205:10 and we're going to take it over to the here and paste this in. And there we go.
205:14 Act as corner cleaning WhatsApp support and lead generation agent. Engage with
205:17 potential customers on WhatsApp to provide potential information about our
205:21 cleaning services. Answer FAQs. answer off instant quotes ra when pricing inquiries arise use the
205:28 instant quote generator tool tools you have this tool and this tool examples
205:33 I'm just going to cut that out for now and then notes ra so that should be all
205:37 good we can start to give this a spin here I am going to zoom out a bit right
205:39 all right so I'm just going to publish this and make sure that everything is
205:44 baked in the second and now we can chat to it here um hey how's it
205:51 going actually slide this across Um, I want to know where you guys are
205:57 located. There you go. Connor's cleaning is located at XYZ. So, it's obviously
206:00 using the knowledge base correctly and say, "What what services do you
206:06 provide?" And we're not asking about quote or it might try to do it at the
206:13 end. Yes. So, see, it's asking if you have any specific requirements, need a
206:16 quote, just let me know. Yeah, sure. I'd like like a quote. Boom. I need the
206:21 property type and square footage. It's a house and it's 1,00 square ft. So now
206:26 the agent is trying to trigger that tool by taking the house and taking the 1,00
206:30 and then putting them into the relevance tool based off what the schema has told
206:33 it how to use the API. It's going to go grab that from relevance, send it back
206:37 to us and there we go. Here are the quotes for you. Ra, if you're interested
206:41 in any specific further service and need assistant, just let me know. I can also
206:44 help you with booking. Now here I would probably change the prompt and make it a
206:48 bit more forceful and say send me like like let's go to the next step right
206:51 now. But for now it's good enough. Um we can say uh sure I'd like to book a deep
206:58 clean. Now it should ask me for my lead information. Okay. Huge
207:04 Jackman is the name is the phone and huge Jack Jackman is email.
207:12 We should be able to see if we go back to our handy dandy air table
207:31 Oh, bang. And huge Jackman is in the CRM here. It does say that it's booked. I
207:34 would play around with the prompt a little bit more to be like, hey, look,
207:37 this is just setting up the next step for someone to call them and book in the
207:40 service. But you can also do appointment setting through agentive as well. Again,
207:42 like I said, it's a little bit more advanced than what we want to do here.
207:47 But as you can see, this is a very different way of approaching building
207:50 agents because you tell it, you basically provide all of the ingredients
207:52 and you use that kind of chef's approach. The knowledge and the tools
207:55 and you connect it all up and you make sure the tools have well described
207:58 schemas so they know how to use it and they know when to trigger them. The
208:01 knowledge base has been included in the prompt and also the tools as well have
208:04 been included in the prompt um telling it how and when to use it. It's really a
208:07 much faster way of building agents from the highle prompting and then people are
208:11 just asking and having sort of a free flowing conversation with it. Okay. And
208:14 just quickly before we go to the step of putting it onto WhatsApp which won't
208:17 take long at all. I do want to show you how you can debug and when you're
208:20 working in agentive um it's helpful to know when tools are being triggered and
208:24 why. So for example, if we go into the transcripts here and we look at this big
208:27 transcript here with 14 messages that we just had. Hey, how's it going? Ra. We
208:33 can see here it's using the tools and we can hover over it and we can see it's
208:36 calling the tool with the URL. It's a post method and we can see the data
208:41 here. I'll just zoom in on that. The property type and the square footage
208:44 that was sent away to relevance are here. So if you're having issues with
208:46 your tools or it's giving weird responses, um you can either come in
208:49 here to the transcripts after the fact. So say maybe this is on WhatsApp um and
208:54 something's going wrong or customers are getting upset. You can come into the
208:57 transcripts here and pick through and see what's going wrong with the tools.
209:00 And just like down here as well, the lead capture, we can see the name,
209:03 email, and phone were all put into this request and sent away to relevance AI.
209:07 And then onto Air Table as a second step. Um, and you can also see the
209:10 output as well. So the output of the tool is all in here. It's basically just
209:14 giving us a confirmation back from Air Table that, yep, everything went well.
209:19 And up here, you can see the output as uh the response with the deep
209:23 cleaning estimates and stuff like that. You can see it a lot more easily if we
209:32 Okay. And if you give this a second once it's finished generating agentive will
209:36 then pop up this show usage and bang there in the editor here. You can then
209:39 debug. Okay. How many tokens are being used? How much is this costing? What was
209:43 the model? Um etc. And then you can see the tools input here. Apartment 500 ft
209:47 etc. And the output as well. So it's really easy to debug those tools while
209:50 you're in Agent. Let's make sure that we've published this. I'm going to publish it again. In
209:55 Agent, we do have version history. So, if you do publish it and you want to
209:58 roll back or look at how you had it set up previously, you can now see that I've
210:02 got two versions, V1 here, and I just took away this little full stop here and
210:06 you can see that that's I've changed the prompt. So you can update it over time.
210:09 You can make edits within Agentive here and test test test. And then when you're
210:12 ready to push that to production and basically if we had this on a WhatsApp
210:16 agent and say I published this, connected to WhatsApp and it was working
210:19 and then I looked through the the transcripts and something wasn't quite
210:22 how I liked it, I could come in here and make edits and then test test and then
210:26 when I was ready to publish it, I click publish and then it's going to push
210:28 those live to the agent. So you're not going to mess things up by playing
210:31 around with things on here. So the final step is of course to deploy it to
210:38 Go to the deploy tab here. You can then click connect WhatsApp. I'm going to click
210:45 continue, get started. You will of course need a Facebook business manager
210:48 to set up this integration fully. That's free with every Facebook account. So, if
210:51 you haven't got one already, I'll leave a link in the description so that you
210:54 can set it up. Takes a few clicks. Then you will see this page here. And you can
210:56 select the business manager you've created. In this case, I'll be using
210:59 this testing one. And then you'll be able to set up a new WhatsApp business
211:02 account, which I can click here. I'll go next. set up a business account
211:09 name. And then this is the display name for the business. And we're going to
211:14 call this a a retail business. Now, you need to provide the phone number that
211:17 you want to connect your agent to. Um, unfortunately, you can't have your own
211:20 personal WhatsApp number and also have a business account running through it. So,
211:23 you need to either buy another SIM card or borrow a friend's number who doesn't
211:26 have WhatsApp, etc. In this case, I'll be using a spare number that I have.
211:29 Then, they're going to send you a verification code to your number, which
211:32 you have to enter in. And then you should see the screen once you've
211:35 successfully passed that verification. So when we continue, it's verifying our
211:39 information for a second. And now our agent is connected to that phone number
211:42 and we're ready to give it a test. So if we go finish here, there's one more
211:45 thing that we need to do on Agentive, which is to click this. Yours may say
211:48 not registered. Don't worry, you can just click this check box here and click
211:53 confirm. Give it a second to connect. Now we've successfully connected our
211:56 agent to that WhatsApp number. Now thing here is this interval. If you're not
211:59 sure what the interval is, you can read this tool tip here. And if you're done
212:02 with the deployment and you want to remove it from that number, you can
212:04 always come back and click deactivate deployment here. But all that's left to
212:07 do now is to test our functionality. Right. So I have it connected to my
212:10 phone here. So I'm just going to show you a little bit of a on screen here of
212:13 me creating this contact and having a message with it. So the number that I
212:17 set up, I can create a new contact and Hey, and you can see on screen here it
212:26 says this is the business using a secure service from Meta. So, this means this
212:29 is a business account um as we've connected it through our WhatsApp uh
212:33 business profile that we set up before. And there we go. We get a
212:36 message back. Hello, thank you for sharing your information. How can I
212:39 assist you today? Um if you have any questions about cleaning service or need
212:42 a quote, feel free to let me know. So, I can say um yes, a quote. Let's ask a
212:47 question to the knowledge base. Where based? There we go. We are based in the
212:55 greater Boston area. It's giving me the uh the correct location there. So, we
212:58 can go for the lead capture now. So, if quote. So, property type uh it's a house
213:12 that is ft. There we go. We're getting the uh estimations and our quote back. Um, it's
213:24 asking if we're interested in any of these services. I'd say yes, I'd like
213:33 please. Now, it should ask me for my contact information. There we go. So, Liam, I
213:41 mean, Liam at mail. Cool. And then we should see it appear over here on our Smith Solar
213:49 CRM. And boom, there it is. So, we've got everything done. That is just one
213:52 run through of using this WhatsApp agent. But as you can see, uh the the
213:56 messages don't come back instantly. So it it feels like it is like it's
213:59 actually could be a real human there applying and it's giving just clear
214:03 information right through WhatsApp. Imagine you are reaching out to maybe
214:06 book an accommodation or you're reaching out to a a cleaning service like this or
214:09 you're reaching out to any kind of business and you want some real
214:11 information directly from what feels like a person. And then you also have
214:15 the functions of getting a real quote of I mean a great use case for this kind of
214:19 thing is like barbers. Like I say, if you you message a barber on WhatsApp,
214:22 maybe you're in in Europe somewhere, you're in South America or you're in
214:25 Central America or and and you want to go to a barber and this is a common
214:28 issue that I've run into when I'm traveling. It's like, I want to message
214:32 this barber, but I might not speak the language that well. And then if you
214:34 message them in English, it will be able to handle that in in English as well as
214:38 in Spanish or in Portuguese or wherever you are. So, this kind of functionality
214:40 built through WhatsApp is a really really great use case for you guys to
214:43 pick up, which is why I wanted to teach you guys it. And we can also go back to
214:47 agentive here. And if we go to our transcripts for this agent, we can see
214:51 the one for today is here. So 12 messages. You can go through the entire
214:54 transcript and you can see it's calling the quote tool here. We see all the
214:57 information that went in and out of it. And then we see the air table lead
215:00 capture information as well. Input and output. So that is how you use a genty
215:04 my software for building these WhatsApp based and also other deployments as
215:07 well. So if we go to studio and we go to deploy, we have Instagram as well. So
215:10 via our mini chat template. You can hook into Instagram and do appointment
215:13 settings and things on Instagram. You can go through Messenger if you want to
215:17 run some Facebook lead ads to Messenger through voice flow as well. Telegram,
215:20 Discord, we have integrations with everything you need as well. So that's
215:23 the end of this build. I hope you enjoyed and uh this is a super handy use
215:27 case um and and deployment really for agents. So now that you understand how
215:36 AI agents works and can build them for yourself, let's talk about the most
215:39 important part of this, which is actually making money with these skills.
215:43 But first, we need to destroy a huge misconception and that you don't need to
215:47 build the next chat GBT or create some revolutionary AI startup in order to
215:51 make money in the AI space. The real opportunity is much much simpler. It's
215:55 just helping businesses to understand and implement AI. This is how I
215:59 monetized my AI agent skills and it has been the most explosive growth I've ever
216:03 experienced in my career. And the good news is, if you've made it this far in
216:06 the video, you are so much closer to being able to tap into this starving
216:10 market for AI services than you think. But don't take my word for it. I'm just
216:13 some guy on the internet after all. Maybe you should listen to some of the
216:15 world's most famous businessmen saying that this is the opportunity to get into
216:19 right now. If I was 25 years old today in 2024, what would I do? What's a good
216:24 sector to get involved in? What business would I get involved in? I think
216:28 everything is looking at AI now in a different way. And I think AI growth is
216:32 going to be exponential. So, anything to do with AI now, what could that be? In
216:36 the simplest form is helping people use the technology. there's going to be a
216:40 massive amount of people wanting to use it that don't know how to and they're
216:44 willing to pay to solve that pain point. So, is that consulting? Not really. It's
216:50 implementation and execution. And so, helping a business do that transfer into
216:54 a world where they're controlling their data and getting information from it.
216:58 Now, the majority of businesses in America, for example, are between 5 and
217:02 500 employees. So, they're small businesses. They create 62% of the jobs.
217:07 They want to use AI. you should help them solve for that and they'll pay you.
217:11 Even another shark, Mark Cuban, is saying the exact same thing that the
217:14 biggest opportunity right now is helping these small to mediumsiz businesses who
217:19 don't understand AI yet, but desperately need it to keep up. And they're
217:22 absolutely right. If we look at the data, it's pretty obvious. According to
217:26 recent data, there's 1.7 million businesses in the US alone that are
217:30 making between $500,000 and $10 million per year. These are small businesses,
217:35 which, as Kevin Oer says, make up 62% of the jobs in the USA. They create 62% of
217:40 the jobs. They want to use AI. You should help them solve for that and
217:43 they'll pay you. These businesses know they need AI to stay competitive, but
217:46 they don't have the time to learn it themselves. And there's basically no one
217:50 there to help them. All of the big consulting firms are looking at other
217:53 big businesses and just leaving these smaller businesses completely ignored,
217:57 but they still make lots of money and they still have a lot of money to invest
218:00 in these kinds of services. Basically, all small businesses are starving for
218:03 some kind of AI services, either education services to help them
218:06 understand what AI is in the first place and why they might need it. There's the
218:10 huge need for consulting services where you help them to identify where AI can
218:13 help with them most in their particular business. And of course, there's
218:16 implementation services where you help them to build and maintain the AI
218:20 systems like the AI agents we've just built. And right now, based on the data
218:23 collected in my community, and we are the largest AI business community on the
218:26 planet right now, for every person or agency that is currently offering AI
218:30 services, there are over 1,100 businesses in the USA alone that need
218:35 help. So, that's a 1 to 1,100 ratio, which means this is a completely
218:38 untapped market. And that's where people like you and I come in, helping these
218:41 hardworking small business owners to understand AI and implement it so that
218:45 they have a chance to keep up. And that's really what drives me and the
218:48 team at Morningside because our company vision is a world where the benefits of
218:52 generative AI are distributed as fairly as possible and they make it to people
218:56 like me and you and the small business owners rather than just all going to
218:58 these giants at the top. And this whole concept of selling services around an
219:02 emerging technology is nothing new. And we saw the exact same pattern when the
219:05 internet first came out. Companies that helped businesses to adapt to the web
219:09 and sort of get online made fortunes. You know, agency.com, Razerfish, etc.
219:13 And I spotted this opportunity within the AI space in 2023 when it wasn't
219:17 anywhere near as clear as it was now. No one really knew how to make money out of
219:20 this stuff. And then I started Morning Side AI. And since then, we've generated
219:24 over $5 million in selling these kinds of AI services of education, consulting,
219:28 and implementation. And we're literally still only just getting started. And the
219:31 best part out of all of this is that as we've proved in this video already, you
219:35 don't need to be a technical genius to understand AI and even to build AI
219:39 agents. You just need to be one step ahead of the businesses that you're
219:42 going to be helping. So, let me show you the three specific ways that you can
219:47 start making money with your AI agent skills. So, as I said, there's basically
219:51 three types of services that you can provide to businesses in order to
219:54 monetize your skills. Firstly, there's education, and this is teaching
219:58 businesses about AI, running workshops, and doing presentations, training their
220:01 teams, and creating courses for them to watch. Businesses are desperate for
220:05 someone who can explain what this stuff is in simple terms and more importantly
220:08 what it can do for them. After watching this video and probably my other huge
220:12 video that I did on AI agents, which I'll link down below, um you will know
220:16 more than enough to start educating businesses on AI and AI agents. Secondly
220:20 is consulting. And this is where you analyze a business's operations and you
220:23 show them where AI can help them save time or make more money. You're
220:27 essentially being their AI strategist. For example, you could go into a
220:30 business and then recommend something like the sales co-pilot system that we
220:34 just made in order to help their struggling sales department. And third
220:37 is implementation. So this is where you actually build and deploy AI solutions
220:41 for businesses. Or better yet, like my agency, you can do all three of these,
220:45 but it did take us 2 years to get here. So there is really no rush. You just
220:48 pick where you want to enter and work your way up to doing more and more if it
220:51 makes sense. Believe it or not, there are people with only a few months
220:54 experience in the AI space selling all of these right now. And the demand from
220:58 businesses is increasing insanely fast right now. I we're seeing this at
221:00 Morningside. Just so many more businesses reaching out. But here's the
221:04 thing. You have one small problem, and that's that you don't quite know enough
221:08 to start moving on this. You are close, but you're not quite there. The way to
221:11 make money in the AI space or with any services really is to create a knowledge
221:15 gap between yourself and the people that you're helping. Your knowledge gap is
221:18 your money maker, and businesses will pay you in proportion to how much more
221:22 you know about AI agents and their business applications than they do. Now,
221:26 while this video has taught you a lot, your knowledge gap is still small. But
221:30 we can fix that. So, let me break down exactly what you need to do next in order to extend
221:35 your knowledge gap to the point where you can start making money. We can call
221:39 this video as step one. So, as long as you've taken notes and followed all the
221:42 tutorials and built the agents alongside me, you're already ahead of most people
221:46 who have no idea about what agents are, how they work, or how to build them. So,
221:49 it's a big step forward with this video. But step two is building even more
221:54 experience building AI agents. So you are more familiar with the platforms and
221:57 better understand the different ways that they can be used to deliver
222:01 different kinds of AI agent use cases or even just AI tools in general. I've only
222:04 really given you a taster here, but I tried to make it as as diverse as
222:07 possible as you could probably tell. In order to do this second step of
222:10 extending your knowledge gap further and building more experience, you can go to
222:14 my free course on school where you'll be able to build another 5 to 10 agents
222:17 following the tutorials that are in there for you. So the link to join my
222:20 free school will be in the description. So, if you blast through all those
222:22 tutorials in there, this is going to further expand your knowledge gap. And
222:26 remember that the more that you know compared to the businesses that you're
222:28 trying to help, the more they're going to pay you. So, step two is building a
222:32 few more agents out and getting a bit more experience on the tools, seeing
222:35 different use cases, etc. And once you've done that, you'll have what I
222:38 call foundational knowledge. So, you understand the core AI concepts that
222:42 we've been through in this video. You can build basic solutions on these
222:45 platforms. You know what's possible for businesses with agents right now. And
222:49 then comes the big decision. Do you want to go deeper technically on this
222:52 building side of getting your hands dirty or do you want to start monetizing
222:56 what you already know? As we've covered, the building and implementing of the AI
222:59 systems is only one of the services that you can sell. So naturally, the
223:03 technical skills needed in order to make money in the implementation services.
223:07 Actually building these systems and businesses is much more greater than
223:11 just having a foundation. But with a good foundation, you're basically ready
223:14 to start having a crack at the other two services of AI education and AI
223:18 consulting. So, the decision of what to do next comes down to really knowing who
223:21 you are and what you are really interested in. And this sounds all woo
223:24 woo and like, oh, you got to know yourself and stuff, but this is I mean
223:29 it very very seriously in that if I use myself an example, I've always loved
223:32 making things, right? I used to build block houses with kids. I used to like
223:36 brew beer with my grandpa. I've always loved tinkering with engines. So when I
223:39 hit this foundational level that you guys will be at after completing those
223:42 extra builds in the free course, I kind of naturally just dove deeper into the
223:45 technical side into building more stuff. I I kept building more and more complex
223:49 AI systems and building upon those skills that I've I've built already,
223:52 which led me ultimately to starting Morningside AI where our first service
223:56 was building AI solutions and systems for clients. But here's the thing, of
223:59 course, a lot of people aren't like me. They don't get as much of a buzz out of
224:03 building things. many of you are going to be much better or enjoy more the
224:06 teaching aspect or working with people and doing the consulting aspect rather
224:10 than building stuff. So in these cases using the foundational knowledge that
224:13 you're going to build up to sell AI education to businesses or AI consulting
224:17 makes a lot more sense. Goes back to the whole Einstein thing about like judging
224:21 a fish on its ability to climb a tree. If to you the building is like a tree
224:23 and you feel like a fish and it's not a really good fit, then there's better
224:26 stuff that you can do and you can find a way to make money in the AI space that
224:30 leans more into your strengths. like in the case of a fish would be swimming,
224:33 right? So, by being honest with yourself and saying, "Hey, look, that's not
224:36 really me. Yeah, sure, I did get it done. I know how that works now, but I
224:39 don't feel any kind of attraction to doing more of that." While you may see
224:42 that as a negative and saying like, "Oh, I don't have what it takes." It's
224:45 actually can be very empowering if you say, "Bang, I'm stopping it here. I'm
224:48 stopping the learning. I'm stopping the procrastination. Now, I'm going straight
224:51 into actually monetizing." It's basically putting a stop on when you do
224:55 this learning big long phase and saying, "No, action starts now. I'm not I'm
224:59 never going to get there, but with this base, I can do a lot and I'm going to
225:02 start taking action with it and making money with it today. So, this
225:05 self-reflection is really what prevents you from getting stuck in an endless
225:08 learning phase of procrastination when you could be out there making money. So,
225:11 in summary, the two routes you have and the two options you have from here are
225:15 if you love building and you kind of naturally feel like you want to learn
225:19 more like like myself when I was at your stage, then just keep going. go and
225:22 watch the free course tutorials on my school and then start going and building
225:25 your own projects and ones for friends and family and whatever you you sort of
225:29 pulled towards naturally and within two to three months you'll have enough
225:32 skills and experience to actually start selling implementation properly. But on
225:35 the other hand, if you haven't fallen in love with the building aspect, then it's
225:39 probably best that you just go go into the free course, smash out the rest of
225:42 those tutorials and finish your foundation and then just get started on
225:45 monetizing your skills either through selling AI education or through selling
225:50 AI consulting. So once you're clear on what kind of AI services you want to sell, getting your
225:55 first few clients is actually pretty straightforward. People try to over
225:58 complicate it, but there's really just two main ways that I'd recommend you do
226:01 this based off all the success I've seen over thousand thousands of people across
226:04 my free and paid communities. The first and by far the easiest method is through
226:07 your warm connections or warm contacts. This means reaching out to people that
226:09 you already have some kind of relationship with, whether it's friends
226:13 or family or kind of acquaintances or even friends of friends that you've met
226:16 once kind of thing. All of these people count as warm connections. So instead of
226:19 trying to convince complete strangers to trust you with your business, you can
226:22 start with people that you already know or have some previous relationship with
226:25 and therefore have an increased level of trust with you through your
226:28 relationship. And I've covered this many, many times on the channel before.
226:31 So on the school post for this video, I will add in my complete guides for warm
226:35 outreach, including resources directly from my AAA accelerator program. The
226:38 second way is using what I call the community content flywheel. So this is
226:42 how you can build long-term momentum beyond just warm outreach. So here's how
226:46 it works. You join the free school community. you start making content
226:48 about what you're learning at each stage. This could be through YouTube
226:51 tutorials, which I mean that worked for me. LinkedIn post is another one um or
226:55 whatever platform you really prefer to create content on. But here's the key.
226:59 You share that content back into the community. So with over 120,000 members,
227:05 by posting it into the community, you get an instant audience and people who
227:08 are really interested in the stuff that you're talking about. So, a perfect
227:11 example of this is a guy called Rory Ridges, a a young guy from the UK who
227:14 joined my free community and basically followed this exact process that I've
227:17 told you in this video so far. So, he took my free course, he learned all the
227:21 basics, built his foundation, then he started posting simple tutorials on
227:25 relevance AI, which you've used in this video already, and he literally just
227:28 started sharing what he'd learned from my videos and making other videos about
227:31 it. And at the start, he was literally just sharing what he'd learned from my
227:34 videos. So, he'd watch a video, then go and kind of make his own video on the
227:37 same sort of topic. And every time he made a tutorial, he'd then share it into
227:41 the community and the community would watch it. They'd give him feedback, go
227:44 and subscribe to his channel. This not only helped him grow faster on YouTube,
227:47 but it also started to position himself as an expert in the community. And he's
227:50 also building his authority in the AI agency space by getting more momentum on
227:54 YouTube. Now, his YouTube channel brings him in enough leads to support his
227:57 growing agency. And I've just seen him recently in the community saying he's
228:00 hiring. So, that's usually a bloody good sign that he's making some good money
228:02 off the back of it. He basically started the same flywheel that took me from zero
228:07 to where I am now. Over $5 million in revenue generated across all my
228:11 businesses and $450,000 plus subscribers in just two years. And so the community
228:15 gives you an audience and the content gives you credibility and together this
228:18 method brings you clients. In the resources for this video on school, I
228:22 will leave links to my complete guide for creating content to generate leads
228:25 just like Rory and I have done successfully. And of course, I'll
228:28 include a link to Rory's channel in the resources in the school community. Now,
228:30 the really important thing to notice with both of these methods we've just
228:34 talked about, the warm outreach and the uh community content flywheel is that
228:37 both of these methods start with giving value first. Whether it's helping your
228:41 warm connections to understand AI or sharing your knowledge through content,
228:47 you have to give before you start to get. Now, I know all of this businessy
228:50 stuff may feel a little bit overwhelming or out of reach for some of you, but
228:55 you'll seriously be amazed at what baby steps add up to in the AI space. you've
228:58 already taken the first step by watching this and following through to the end of
229:02 the video. So, congratulations on that and I seriously seriously want to give
229:05 you a pat on the back and you should give one to yourself. But all you need
229:09 to do from now is to keep this momentum going. The next step for all of you is
229:13 pretty damn clear. You need to jump in the free community. That's by far the
229:16 best place if you're moderately interested in doing any of this stuff.
229:19 My community is the number one place to go. It is 100% free to get into. And
229:21 once you're in there, drop an introduction post saying who you are,
229:24 what you're about, why you want to do this. Then start working your way
229:27 through my free course material. I've poured everything I've learned about AI
229:31 and AI business into videos like these. And on the free course, they're all
229:33 there in a nice sequence for you to work through on school. And each time you
229:37 complete a video and you complete the tutorial, you can click the little check
229:40 box so that you can keep stacking those small wins and those baby steps towards
229:43 AI literacy and succeeding in the space. All of the resources mentioned in the
229:46 selling part of this video will be on the school post for this video. So you
229:49 go into school, you go to the YouTube resources tab, and then this video will
229:52 be right there. So don't forget to check all those out. And of course, all of the
229:56 resources for the tutorials, if you haven't already done them, are included
229:59 on those posts as well. And finally, if you made it this far, could you please
230:02 do me a favor, leave a like on the video, and drop a comment down below.
230:05 Let me know what you like the most, what you want to see more of. Click the share
230:08 button, send it to your friends and family and loved ones so they can start
230:11 learning these skills, too. All of these actions really help my video to reach
230:13 more people in the YouTube algorithm. And if you subscribe to the channel,
230:16 you'll be able to see a lot more content like this helping you to understand AI
230:20 and more so how to build businesses around this incredible opportunity. And
230:23 if you're still hungry for more and you want to watch my complete guide to
230:26 building an AI business, that's going to be linked up here. But that is all for
230:30 the video. I'm so excited for you to get cracking on this. I sincerely hope
230:33 you've got something out of this because I put a lot of work into it as did my
230:36 team. So I'm just really really wishing you all all the all all the best. I'll
