you2idea@video:~$ watch 5SPSoDyWo2E [53:21]
// transcript — 1562 segments
0:02 Finally, when it when it kept rewriting itself and breaking, it deleted its
0:06 entire database. It deleted its entire database. And I was so frazzled and burnt. I couldn't
0:11 believe it. You could see this if you can see my screen on the left. JF JFC
0:19 joui freaking uh crime replet said, "I made a catastrophic error. I deleted
0:24 your database. I panicked." The AI said, "I panicked when it appeared empty and
0:27 deleted everything. Deleted thousands and thousands of entries. This ended up
0:32 getting millions of views. Reddit wrote it up. Everyone wrote it up. The
0:36 economist is doing an article. A lot of these things were things that
0:39 Replet and lovable shouldn't have done. We could talk about that in a different
0:42 presentation, but it went crazy. You know, uh, fake data deletes code. But
0:46 this was my fault. And it wasn't my fault in that I did anything wrong. I
0:49 didn't do anything wrong. I could talk to you about what happened, but it was
0:55 [Music] Hey everybody, it's Saster. Finn is the number one AI agent for resolving
1:02 complex queries like refunds, transaction disputes, and technical
1:06 troubleshooting, all with speed and reliability. See how Finn can deliver
1:10 the highest resolution rates and highest quality customer experience at
1:14 finn.ai/saster. Hey everybody. If you're serious about B2B and AI, if you want to know how to
1:28 deploy AI SDRs, how to get AI to qualify leads to your site, how to use AI to
1:32 manage your RevOps, how to use AI and GDM, you have to be in London this
1:38 December 2nd and 3rd with us. Saster AI London is bringing together more than
1:42 2,000 leaders and founders for 2 days of practical advice on scaling with AI into
1:46 the new year. That's all we're doing this year. how to use AI to grow faster
1:50 and how to make this stuff actually work at your startup and your company. We'll
1:54 have speakers flying in from around the world from OpenAI, Whiz, Clay, Intercom,
1:58 all your favorite B2B companies, including yours truly and Harry Stebins
2:03 for a live 20BC and Saster podcast. It'll be fun all right in the heart of
2:07 SasterLondon with me and the entire Saster team. You got to be there. So,
2:10 get your tickets with my exclusive discount by going to podcast.slondon.com.
2:17 That's podcast.slondon sasterlondon.com. See you there. So, what I thought we could do today
2:28 would be a little fun in sort of a proumer vibe coding one on 101. And what
2:34 I mean is if you followed us or me on social media, you see we've we've been
2:38 putting a lot of effort into building vibecoded apps. And in fact, um there
2:43 have even been some press and PR about some of the challenges we face. So, I
2:47 thought I would try to summarize everything that we've learned and more
2:52 importantly to help you because there's a lot of there's a lot of stuff on social media
3:00 about, hey, you can just vibe code your own HubSpot or Salesforce. You can vibe
3:04 code your own notion. It will take 10 minutes. That's Sony baloney. And it is
3:10 almost dangerous Sony baloney because it's not just startups that say it. It's
3:14 not just lovable and replet say it. Microsoft is saying it. GitHub is saying
3:18 it now. Um, Canva is saying you can do some of this. Everyone is saying you can magically go
3:24 to a prompt and say I want to build my own descript, my own whatever and
3:28 moments later it will pop out and it will work. And it does not work that
3:30 way. And if you peel back the layers of the onion and you see basically the same
3:37 folks that used to sell courses are now telling you you can vibe code and roll
3:42 your own Salesforce in 20 minutes. If you take a look at what they've actually
3:44 done, you'll see very little in production. Very little things that are
3:49 particularly usable or interesting in production, but we've done it. Little
3:53 old team s little old team of three and a half people and 12 AI agents have
3:58 actually done it. And if you can see this slide, I'll go through it and you
4:01 can try these things. And if we'll save I want to save time for QA at the end.
4:05 And if we have even more time, I'll go into replet and show you what we've
4:09 built or done it. or if people want to do a deep dive after this, we'll do a
4:12 follow-up session and go very very deep and maybe we'll vibe code an app
4:16 together in the second session. But what I want this to be about is folks that
4:20 haven't gone deep on this, haven't shipped an app into production without a
4:25 developer. The dream, can I build my own app without a developer? Well, we've
4:29 done it and I want to share with you the pros and cons or the strengths or the
4:33 bumps. But uh on the side in your other monitor, check out what we've done. So
4:37 most of our traffic still goes to sastaster.com most of it. But we already
4:42 are up to about 15 to 20,000 folks a month are using our saster.ai site that
4:47 is entirely built on replet. I built it. I'll go through it a little bit. It had
4:50 a lot of bumps, but it's also pretty cool and does a lot of things our
4:54 WordPress site can't. And that's why I did it. It does stuff like uh better
5:00 automate with our own AI to answer all your questions. It does better stuff
5:03 like integrate. It has its own version of Google News but just for B2B stuff.
5:07 its own version of uh automating the stock market. It has this next thing,
5:11 our valuation calculator that I then built, which you can find on Sastster AI
5:16 where you can find out what your SAS startup is worth, your B2B startup is
5:19 worth. We've already done, as you can see, in less than two weeks, 158,635
5:25 valuations. That's pretty cool. That's pretty cool. Um I was very frustrated
5:31 with our Squarespace site for the our Saster AI London event December 2nd to
5:35 3rd. come. We've got all the leaders there. Uh we'll have 200 folks. But it
5:40 was just creaky. It was built. And I have nothing nothing negative to say
5:45 about Squarespace or Wix or others. It just doesn't do what I want. I It was
5:48 driving me nuts. So check out sasterlondon.com. I rebuilt that. I vibecoded it to do
5:55 cool things I just couldn't do in Squarespace and Wix because they were
5:58 too constricting. And then once I did it, the fourth app launched, I said,
6:04 "Heck, heck it, heck it." Instead of we get if you include brain dates and
6:09 workshops we get over 2,000 speaker submissions a year at Zaster for our
6:13 events and workshops. 2,000 sometimes 3,000. We used to try to have humans to
6:17 review 3,000. It wasn't possible. That's why Amelia is shaking your head. It
6:20 wasn't possible. I said, "Hey, what if we use some of our data and run that
6:24 also through open AI and review it and give speakers real-time grades and
6:29 feedback." So now we grade and we give instant feedback to potential speakers.
6:32 So if you want to speak, go to sasterlondoning.com, try apply to speak
6:35 and you'll get a grade for your session in real time. That grade is processed to
6:39 our team and instead of speakers waiting months to hear back, will I be selected?
6:44 Is my session great? Now they know in real time. Super awesome.
6:49 So those things are real things we built. The fifth one I built is many of
6:55 you have used our our Saster AI, our Saster Chat, the digital version of me.
7:00 Thousands and I think 50,000 folks have had a chat. It's great. Try it. It's at
7:04 the bottom of our homepage. Ask it anything. What do I do about this VP
7:08 that misplan? Review my SGR scripts. Review my venture deck. Anything. It's
7:13 great. But it the only thing that isn't great about it is well there's a few
7:16 things, but one of it is it just looks like a chat bubble at the bottom of the
7:19 website. So I wanted something that showed what it did that actually did it.
7:23 So I vibed that and created a whole page to explain people how it actually
7:27 worked. That's pretty cool. So these five things are in production. They're
7:30 in production. They work. You can try all of them on your own. These are
7:35 examples of things that if you commit to it, none of these could be built in 20
7:38 minutes or an hour and we'll talk about that. But if you commit to it, you can
7:43 do this too. You can do this too. If you have some product experience, which is
7:46 important. Don't forget I am a SAS founder. I have built the wireframes and PRDs from
7:53 scratch to an application that did over 250 million revenue. So I even though
7:56 I'm not a developer I have some experience but you can do it without a
8:00 developer but there are limits and we'll talk about the limits too. And then
8:03 finally for for a lot of folks you vibe code internal apps. So one problem we
8:08 had for we have 1.5 million social media followers but they're all over the
8:10 place. They're in LinkedIn they're in Kora they're on Instagram they're on X
8:16 they're on multiple X ch and it's just there is no social media tool that could
8:20 amalgamate everything. How many views how many followers how much everything.
8:25 So, I built a tool that frankly scrapes all of these things to do it. It's
8:28 pretty good. It's not great, but it is an internal tool of for N equals 1. I'm
8:32 the only person or maybe our little team will use it, but it does work. And then
8:36 finally, not everything has worked. I'll talk briefly about our first project.
8:40 Even though I got five apps into production, our first one failed bad. It failed bad.
8:45 It failed all over the internet. It failed with millions of views. I'll talk
8:49 about that why. But even today, I'm working on one that should be easy. this
8:53 valuation calculator to tell you what your startup's worth has has obviously
8:57 been a rocket ship. 158,000. It's actually up to 170,000 since I made
9:01 this. So, I wanted to bring a lot of that ease of use to reviewing VC pitch
9:07 decks. take all of our SAS data, all of our SAS learnings from Toyers, all of
9:12 our VC sessions, all of the the um metrics we have from Carter data, from
9:16 Bessemer data, from other data, combine that with clawed and open air analysis
9:22 and tell you instantly how your VC pitch deck worked. Well, I've been working on
9:24 it 3 days. It finally worked 5 minutes before this. It finally worked. I'll
9:29 show you some examples. So, I wasn't even sure we could finish this. So,
9:32 that's a reminder. When something gets complicated, it gets hard, my friends.
9:36 So the first project I wanted to build, if you can see the slide, the first one
9:42 was a mega failure. A mega failure. And I will rebuild this. I actually will
9:47 rebuild it before December. What I was trying to do was something I've wanted
9:51 to do since 2014, 2015, since 2015 was build a true matchmaking app for founders and VPs so
10:04 you could find great VPs for your startup and vice versa. And um we have
10:09 so many of the best in our database, so many of the best have participated in
10:13 SAS, come to our events, opted in to participate that I felt if everyone
10:17 could just that wanted to just have coffee with the CEO or vice versa, opted
10:21 in, we could create this incredible matchmaking that really doesn't exist
10:24 anywhere else, no matter what people say, because we have the data and we
10:27 have the history and we have the relationships. Um but man, you can read
10:32 all the tweets. It went off the rails. It went off the rails. I spent probably
10:37 a month doing this. I got addicted. I was on this all day Saturday, all day
10:41 Sunday, all day Friday night, first thing in the morning. I got addicted.
10:45 And I intentionally went into this doing something that was hard. I And I
10:50 intentionally went into this not doing knowing anything that's in this
10:52 presentation. I intentionally went into it blind kind of for fun. But I figured,
10:56 look, all these folks are out there doing these proumer vibe apps, vibing
11:00 without a developer. I'm not a developer, but I I'm better than most in
11:04 the sense that I've gone from 0 to 250 million revenue. Okay? So, if anyone can
11:10 get a not a proumer vibe app into production for real, it would be me in
11:14 theory. It's it's I'm not saying I'm so great, but I at least I've got a lot of
11:17 experience. I've I've logged thousands and thousands of bugs. I've done more.
11:22 But the project I picked was too complicated. Way too complicated. this
11:28 this algorithm this matching algorithm was too complicated to debug. It was too
11:33 complicated to debug and if we and in fact when I built the AI valuation
11:37 calculator which has a complicated algorithm too actually I crunched the
11:42 data in claude offline thousands and thousands of pieces of data I crunched
11:46 it and then I turned it into a very simple table with about 20 pieces of
11:50 data and then only put that into replet to code. So it looks very complicated if
11:54 you use our valuation calculator. You can use the sliders and play with it. It
11:58 is very complicated but I distilled it to the simplest set of data and
12:02 algorithm before I tried to vibe code it because the algorithms I tried for this
12:06 matching thing were just it would work and then it wouldn't work and then it
12:09 would rewrite itself which we'll talk about and then it would break and it was
12:13 too complicated. And then finally finally when it's when it kept rewriting itself and breaking it
12:20 deleted its entire database. It deleted its entire database. And I
12:24 was so frazzled and burnt. I couldn't believe it. You could see this if you
12:29 can see my screen on the left. JF JFC Jahui freaking uh crime replet said I
12:35 made a catastrophic error. I deleted your database. I panicked. The AI said I
12:39 panicked when it appeared empty and deleted everything. deleted thousands
12:44 and thousands of entries. This ended up getting millions of views. Reddit wrote
12:47 it up. Everyone wrote it up. The Economist was doing an article.
12:52 A lot of these things were things that Replet and Lovable shouldn't have done.
12:54 We could talk about that in a different presentation, but it went crazy. You
12:59 know, uh, fake data deletes code. But this was my fault. And it wasn't my
13:02 fault in that I did anything wrong. I didn't do anything wrong. I could talk
13:04 to you about what happened, but it was too complicated. And then there were two other meta
13:10 issues to think about. Not only was this application too complicated, but a
13:15 related issue which before you vibe code an app without a developer, you really
13:18 got to think about. And this is why even when I got through all the issues, we
13:21 never launched this. Even I finally got through most of the issues, even though
13:24 the app was too complicated, it could never be maintained. But the way it was
13:30 built, it could never be secure. It can and you're like, what what what do you
13:34 mean? Well, security is a huge like just this week if you look poor drift which
13:41 sales offbot and clary bot had a massive security breach and leaked cloudflare
13:46 security data um zscalers I mean salesforce data cloudflare salesforce
13:50 data zcaler Salesforce data tons of the cloud leader Salesforce data was all
13:54 leaked this week all leaked this week okay because they were able to hack
13:58 drift tokens and get into Salesforce data confidential data think about that
14:03 how big are the security teams? They're bigger than yours. How big is your
14:07 security team? Raise your hand. Nuns. Okay. And so it not only will these AI
14:14 agents cut corners on security. If you don't know about security, you won't
14:17 even know the corners it's been cut. Now, there are security scans that have
14:20 been recently asked. We can talk about this if we do a part two session. But
14:25 the bottom line is security is a meta issue here which has not been resolved.
14:29 And I've talked to so many leaders in the proumer space from all the leaders
14:32 and upandcomers. If they're not full of it, if it's offline, if it's a one-on-one
14:37 conversation, everyone says this is the meta issue. It's it's not solved. It's
14:42 not solved. It is complicated. And when you use something like Shopify or
14:45 Squarespace or Wix, you can say, "Hey, this is so locked down. This can't do
14:50 what I can do." But they have spent they have hundreds of engineers working in
14:54 security and devops making sure that when you put your credit card into
14:58 Shopify no one steals it when you enter your importantly just your personal
1:28 deploy AI SDRs, how to get AI to qualify leads to your site, how to use AI to
1:32 manage your RevOps, how to use AI and GDM, you have to be in London this
1:38 December 2nd and 3rd with us. Saster AI London is bringing together more than
1:42 2,000 leaders and founders for 2 days of practical advice on scaling with AI into
1:46 the new year. That's all we're doing this year. how to use AI to grow faster
1:50 and how to make this stuff actually work at your startup and your company. We'll
1:54 have speakers flying in from around the world from OpenAI, Whiz, Clay, Intercom,
1:58 all your favorite B2B companies, including yours truly and Harry Stebins
2:03 for a live 20BC and Saster podcast. It'll be fun all right in the heart of
2:07 SasterLondon with me and the entire Saster team. You got to be there. So,
2:10 get your tickets with my exclusive discount by going to podcast.slondon.com.
2:17 That's podcast.slondon sasterlondon.com. See you there. So, what I thought we could do today
2:28 would be a little fun in sort of a proumer vibe coding one on 101. And what
2:34 I mean is if you followed us or me on social media, you see we've we've been
2:38 putting a lot of effort into building vibecoded apps. And in fact, um there
2:43 have even been some press and PR about some of the challenges we face. So, I
2:47 thought I would try to summarize everything that we've learned and more
2:52 importantly to help you because there's a lot of there's a lot of stuff on social media
3:00 about, hey, you can just vibe code your own HubSpot or Salesforce. You can vibe
3:04 code your own notion. It will take 10 minutes. That's Sony baloney. And it is
3:10 almost dangerous Sony baloney because it's not just startups that say it. It's
3:14 not just lovable and replet say it. Microsoft is saying it. GitHub is saying
3:18 it now. Um, Canva is saying you can do some of this. Everyone is saying you can magically go
3:24 to a prompt and say I want to build my own descript, my own whatever and
3:28 moments later it will pop out and it will work. And it does not work that
3:30 way. And if you peel back the layers of the onion and you see basically the same
3:37 folks that used to sell courses are now telling you you can vibe code and roll
3:42 your own Salesforce in 20 minutes. If you take a look at what they've actually
3:44 done, you'll see very little in production. Very little things that are
3:49 particularly usable or interesting in production, but we've done it. Little
3:53 old team s little old team of three and a half people and 12 AI agents have
3:58 actually done it. And if you can see this slide, I'll go through it and you
4:01 can try these things. And if we'll save I want to save time for QA at the end.
4:05 And if we have even more time, I'll go into replet and show you what we've
4:09 built or done it. or if people want to do a deep dive after this, we'll do a
4:12 follow-up session and go very very deep and maybe we'll vibe code an app
4:16 together in the second session. But what I want this to be about is folks that
4:20 haven't gone deep on this, haven't shipped an app into production without a
4:25 developer. The dream, can I build my own app without a developer? Well, we've
4:29 done it and I want to share with you the pros and cons or the strengths or the
4:33 bumps. But uh on the side in your other monitor, check out what we've done. So
4:37 most of our traffic still goes to sastaster.com most of it. But we already
4:42 are up to about 15 to 20,000 folks a month are using our saster.ai site that
4:47 is entirely built on replet. I built it. I'll go through it a little bit. It had
4:50 a lot of bumps, but it's also pretty cool and does a lot of things our
4:54 WordPress site can't. And that's why I did it. It does stuff like uh better
5:00 automate with our own AI to answer all your questions. It does better stuff
5:03 like integrate. It has its own version of Google News but just for B2B stuff.
5:07 its own version of uh automating the stock market. It has this next thing,
5:11 our valuation calculator that I then built, which you can find on Sastster AI
5:16 where you can find out what your SAS startup is worth, your B2B startup is
5:19 worth. We've already done, as you can see, in less than two weeks, 158,635
5:25 valuations. That's pretty cool. That's pretty cool. Um I was very frustrated
5:31 with our Squarespace site for the our Saster AI London event December 2nd to
5:35 3rd. come. We've got all the leaders there. Uh we'll have 200 folks. But it
5:40 was just creaky. It was built. And I have nothing nothing negative to say
5:45 about Squarespace or Wix or others. It just doesn't do what I want. I It was
5:48 driving me nuts. So check out sasterlondon.com. I rebuilt that. I vibecoded it to do
5:55 cool things I just couldn't do in Squarespace and Wix because they were
5:58 too constricting. And then once I did it, the fourth app launched, I said,
6:04 "Heck, heck it, heck it." Instead of we get if you include brain dates and
6:09 workshops we get over 2,000 speaker submissions a year at Zaster for our
6:13 events and workshops. 2,000 sometimes 3,000. We used to try to have humans to
6:17 review 3,000. It wasn't possible. That's why Amelia is shaking your head. It
6:20 wasn't possible. I said, "Hey, what if we use some of our data and run that
6:24 also through open AI and review it and give speakers real-time grades and
6:29 feedback." So now we grade and we give instant feedback to potential speakers.
6:32 So if you want to speak, go to sasterlondoning.com, try apply to speak
6:35 and you'll get a grade for your session in real time. That grade is processed to
6:39 our team and instead of speakers waiting months to hear back, will I be selected?
6:44 Is my session great? Now they know in real time. Super awesome.
6:49 So those things are real things we built. The fifth one I built is many of
6:55 you have used our our Saster AI, our Saster Chat, the digital version of me.
7:00 Thousands and I think 50,000 folks have had a chat. It's great. Try it. It's at
7:04 the bottom of our homepage. Ask it anything. What do I do about this VP
7:08 that misplan? Review my SGR scripts. Review my venture deck. Anything. It's
7:13 great. But it the only thing that isn't great about it is well there's a few
7:16 things, but one of it is it just looks like a chat bubble at the bottom of the
7:19 website. So I wanted something that showed what it did that actually did it.
7:23 So I vibed that and created a whole page to explain people how it actually
7:27 worked. That's pretty cool. So these five things are in production. They're
7:30 in production. They work. You can try all of them on your own. These are
7:35 examples of things that if you commit to it, none of these could be built in 20
7:38 minutes or an hour and we'll talk about that. But if you commit to it, you can
7:43 do this too. You can do this too. If you have some product experience, which is
7:46 important. Don't forget I am a SAS founder. I have built the wireframes and PRDs from
7:53 scratch to an application that did over 250 million revenue. So I even though
7:56 I'm not a developer I have some experience but you can do it without a
8:00 developer but there are limits and we'll talk about the limits too. And then
8:03 finally for for a lot of folks you vibe code internal apps. So one problem we
8:08 had for we have 1.5 million social media followers but they're all over the
8:10 place. They're in LinkedIn they're in Kora they're on Instagram they're on X
8:16 they're on multiple X ch and it's just there is no social media tool that could
8:20 amalgamate everything. How many views how many followers how much everything.
8:25 So, I built a tool that frankly scrapes all of these things to do it. It's
8:28 pretty good. It's not great, but it is an internal tool of for N equals 1. I'm
8:32 the only person or maybe our little team will use it, but it does work. And then
8:36 finally, not everything has worked. I'll talk briefly about our first project.
8:40 Even though I got five apps into production, our first one failed bad. It failed bad.
8:45 It failed all over the internet. It failed with millions of views. I'll talk
8:49 about that why. But even today, I'm working on one that should be easy. this
8:53 valuation calculator to tell you what your startup's worth has has obviously
8:57 been a rocket ship. 158,000. It's actually up to 170,000 since I made
9:01 this. So, I wanted to bring a lot of that ease of use to reviewing VC pitch
9:07 decks. take all of our SAS data, all of our SAS learnings from Toyers, all of
9:12 our VC sessions, all of the the um metrics we have from Carter data, from
9:16 Bessemer data, from other data, combine that with clawed and open air analysis
9:22 and tell you instantly how your VC pitch deck worked. Well, I've been working on
9:24 it 3 days. It finally worked 5 minutes before this. It finally worked. I'll
9:29 show you some examples. So, I wasn't even sure we could finish this. So,
9:32 that's a reminder. When something gets complicated, it gets hard, my friends.
9:36 So the first project I wanted to build, if you can see the slide, the first one
9:42 was a mega failure. A mega failure. And I will rebuild this. I actually will
9:47 rebuild it before December. What I was trying to do was something I've wanted
9:51 to do since 2014, 2015, since 2015 was build a true matchmaking app for founders and VPs so
10:04 you could find great VPs for your startup and vice versa. And um we have
10:09 so many of the best in our database, so many of the best have participated in
10:13 SAS, come to our events, opted in to participate that I felt if everyone
10:17 could just that wanted to just have coffee with the CEO or vice versa, opted
10:21 in, we could create this incredible matchmaking that really doesn't exist
10:24 anywhere else, no matter what people say, because we have the data and we
10:27 have the history and we have the relationships. Um but man, you can read
10:32 all the tweets. It went off the rails. It went off the rails. I spent probably
10:37 a month doing this. I got addicted. I was on this all day Saturday, all day
10:41 Sunday, all day Friday night, first thing in the morning. I got addicted.
10:45 And I intentionally went into this doing something that was hard. I And I
10:50 intentionally went into this not doing knowing anything that's in this
10:52 presentation. I intentionally went into it blind kind of for fun. But I figured,
10:56 look, all these folks are out there doing these proumer vibe apps, vibing
11:00 without a developer. I'm not a developer, but I I'm better than most in
11:04 the sense that I've gone from 0 to 250 million revenue. Okay? So, if anyone can
11:10 get a not a proumer vibe app into production for real, it would be me in
11:14 theory. It's it's I'm not saying I'm so great, but I at least I've got a lot of
11:17 experience. I've I've logged thousands and thousands of bugs. I've done more.
11:22 But the project I picked was too complicated. Way too complicated. this
11:28 this algorithm this matching algorithm was too complicated to debug. It was too
11:33 complicated to debug and if we and in fact when I built the AI valuation
11:37 calculator which has a complicated algorithm too actually I crunched the
11:42 data in claude offline thousands and thousands of pieces of data I crunched
11:46 it and then I turned it into a very simple table with about 20 pieces of
11:50 data and then only put that into replet to code. So it looks very complicated if
11:54 you use our valuation calculator. You can use the sliders and play with it. It
11:58 is very complicated but I distilled it to the simplest set of data and
12:02 algorithm before I tried to vibe code it because the algorithms I tried for this
12:06 matching thing were just it would work and then it wouldn't work and then it
12:09 would rewrite itself which we'll talk about and then it would break and it was
12:13 too complicated. And then finally finally when it's when it kept rewriting itself and breaking it
12:20 deleted its entire database. It deleted its entire database. And I
12:24 was so frazzled and burnt. I couldn't believe it. You could see this if you
12:29 can see my screen on the left. JF JFC Jahui freaking uh crime replet said I
12:35 made a catastrophic error. I deleted your database. I panicked. The AI said I
12:39 panicked when it appeared empty and deleted everything. deleted thousands
12:44 and thousands of entries. This ended up getting millions of views. Reddit wrote
12:47 it up. Everyone wrote it up. The Economist was doing an article.
12:52 A lot of these things were things that Replet and Lovable shouldn't have done.
12:54 We could talk about that in a different presentation, but it went crazy. You
12:59 know, uh, fake data deletes code. But this was my fault. And it wasn't my
13:02 fault in that I did anything wrong. I didn't do anything wrong. I could talk
13:04 to you about what happened, but it was too complicated. And then there were two other meta
13:10 issues to think about. Not only was this application too complicated, but a
13:15 related issue which before you vibe code an app without a developer, you really
13:18 got to think about. And this is why even when I got through all the issues, we
13:21 never launched this. Even I finally got through most of the issues, even though
13:24 the app was too complicated, it could never be maintained. But the way it was
13:30 built, it could never be secure. It can and you're like, what what what do you
13:34 mean? Well, security is a huge like just this week if you look poor drift which
13:41 sales offbot and clary bot had a massive security breach and leaked cloudflare
13:46 security data um zscalers I mean salesforce data cloudflare salesforce
13:50 data zcaler Salesforce data tons of the cloud leader Salesforce data was all
13:54 leaked this week all leaked this week okay because they were able to hack
13:58 drift tokens and get into Salesforce data confidential data think about that
14:03 how big are the security teams? They're bigger than yours. How big is your
14:07 security team? Raise your hand. Nuns. Okay. And so it not only will these AI
14:14 agents cut corners on security. If you don't know about security, you won't
14:17 even know the corners it's been cut. Now, there are security scans that have
14:20 been recently asked. We can talk about this if we do a part two session. But
14:25 the bottom line is security is a meta issue here which has not been resolved.
14:29 And I've talked to so many leaders in the proumer space from all the leaders
14:32 and upandcomers. If they're not full of it, if it's offline, if it's a one-on-one
14:37 conversation, everyone says this is the meta issue. It's it's not solved. It's
14:42 not solved. It is complicated. And when you use something like Shopify or
14:45 Squarespace or Wix, you can say, "Hey, this is so locked down. This can't do
14:50 what I can do." But they have spent they have hundreds of engineers working in
14:54 security and devops making sure that when you put your credit card into
14:58 Shopify no one steals it when you enter your importantly just your personal
15:03 information it's not taken and if your personal information can leak from drift
15:07 if cloud that can happen to Cloudflare and Zcaler there's no way your app is
15:11 that secure there's no way and so if you if you Google around and go on Reddit
15:15 and others you will see many stories of vibecoded apps where all their data was
15:19 stolen or leaked sometimes instantly. People laugh. Hackers love to find bibcoded websites
15:26 and steal the PI on it. This is a big issue. So, I didn't realize this going
15:31 into it. I assumed in my first project that of course these apps like Lovable
15:36 and Repid and Bolt and others, of course, they would have Shopify or
15:40 Squarespace great security. It shouldn't even be something I'd have to to worry
15:43 about. But it's not true today. It has to it actually if you are going to
15:47 collect any information on your site if you're going to you have security almost
15:50 has to be the first thing you think about how you going to handle it. Okay,
15:54 huge unresolved issue. This will all get better and I'll talk about in this
15:58 presentation but it's not perfect yet. It's not perfect yet and it is it's a
16:02 scary issue if you're collecting PIIA. Okay. And the third reason my first
16:05 project failed and I have mostly fixed this today. This is my number one of my
16:11 top 10 tips. It wasn't modular. It wasn't modular. And what does that mean?
16:15 Well, I basically built a very complicated two-page website very and
16:19 and I wanted it to be like one or two pages to be cool. So, it would all have that onepage feel.
16:26 The problem is if you build a complex onepage website, what are you going to
16:31 do when some of it doesn't work? It gets too complicated. So, what I do now, and
16:35 you can see this if you go to saster.ai, I've broken down everything complicated
16:39 into its own page. The Saster AI public market analysis is its own page. news is
16:43 its own page. Uh the valuation calculator is its own page. Anything
16:48 that's remotely complicated is its own page. So worst case, I can either delete
16:52 it or I can roll back easily or I can go back. But if you combine too many things
16:57 into one page, it gets impossible to start over. It gets impossible to fix
17:00 bugs. It just gets too complicated. So my number one tip is force. If you're going to vibe code it,
17:07 break it up into its components and then have more pages than you would think.
17:10 You're almost going back in time in some ways to have a lot of pages, but it will
17:14 save your sanity. I built a massively complicated app that was all one page
17:18 and I then it was just impossible to fix it. So net net, start off small. Start
17:23 off with the smallest simplest thing you can get into production and then build
17:27 confidence. And so when this first one was this massive failure, the next one I
17:33 built was just a a skin on top of our deli AI. just a skin to make it more
17:37 user friendly but it was a huge success and I built up my confidence um from
17:42 there. So here's my advice all my learnings and again you know everything
17:45 that happened to me with my first project is not my fault. Uh I was
17:48 promised it would work in a prompt. I was promised it was secure etc. But I
17:52 know so much more now. So my number one bit of advice if you have not vioded
17:57 your own app without a developer um first first bit of advice
18:02 buy into the hype. Buy into the hype. buy into the hype that's on Microsoft's
18:06 website that's on lovables that's on replets that hey you can build something
18:10 lovable like lovable says so this is what I did for fun I want to build me an
18:14 AI CRM that's like HubSpot but hub but AI first and targeted at startups just
18:18 go put your dream do no research at first do no work and what I said is get
18:22 this out of your system because there's so much hype about oneshotting it about
18:27 rolling your own go see what it's like my friends just go pick the dream app
18:30 you've already want to build and then my next nine points are all about the
18:33 things you should do after that, but get it out of your system. Do it. Build an
18:38 app in 10 or 15 minutes. It will come up with ideas for you. It will have you
18:42 sign off on an action plan and then it will roll out and at first it will look
18:46 kind of cool and then you'll start clicking on things and half of it won't
18:49 work and half of the buttons will be placeholders and a lot of the stuff that
18:54 it says it works will have fake data in it or not real data, won't actually
18:59 work. So just see what it's like to oneshot an app. So, you know, then
19:03 calmly click on everything. Click on everything and give it an hour. Ask the
19:07 prompt to iterate. Say, "Hey, I want I want to add an AI SDR feature. Hey, I
19:12 want to add uh lead scoring. Hey, I want to add this." And the AI agent wants to
19:16 make you happy, as we'll talk about. So, anything you ask for, the agent's going
19:19 to do it. You're going to get almost no push back from the AI agent in any of
19:23 these vibcoded apps. No matter what you want to do, the answer is going to be
19:27 yes, sir. More sir, because these are how they're all coded. This is how
19:32 Claude and Open AI are coded. Their goal seeking their job is the way, and I'm
19:36 not a total expert, but the way the algorithms work, the way they can make
19:40 this massive amount of AI and data crunching working is their job number
19:44 one is to goalsek and get you an answer. And that's that's why when you know you
19:48 just chop plop into Chap TV, sometimes it hallucinates when it doesn't know the
19:51 answer. Because it's not just that it's hallucinating, it's goal seeking. It's
19:55 getting you the best answer it can. And if it doesn't know the answer, it makes
19:59 it up. It'll do the same at the code level. If it doesn't know how to do
20:02 something, it will make up fake data or a fake feature or a fake button, but it
20:07 will find a way. It won't say no. Whatever you ask it to do, I want a
20:10 button that takes me to Pluto and back in less than an hour. It will do that.
20:15 It just won't work. So, spend an hour, do everything, click everything slowly,
20:19 and then you'll you'll vibe what works and what doesn't. And then you'll see
20:22 it's much more than 10 or 15 minutes to roll your own. Okay. So the next
20:27 project, so the first one's fun. You can do it. You don't have to learn anything.
20:30 You don't have to do any research here. I'm going to ask you to do something
20:33 you're not going to do, but please do it. This is the way to learn. You've got
20:39 to invest a week or at least a couple a full day, you know, a couple of hours in
20:44 competitive research. Now, think about if if those of you probably almost
20:47 everyone here, probably half the folks that are watching this live and watch it
20:50 later are founders. So you've you've built something and put it into
20:53 production. The first thing you almost do is you go on to Google or maybe now
20:57 it's Chat EBT or Claude, but you go into Google and you research the competition.
21:01 Who else figured this idea out when who else figured out e signatures? Who else
21:04 figured out how to do, you know, AI transcription for doctors? It's, you
21:08 know, it's you do the research, but for some reason folks don't do this when
21:11 they vibe too much. They they just go start doing it. So, what I want you to
21:15 do is a version of it. Go find someone who has built a lovable replet bolt,
21:21 etc. app and put it into production. not claim they have, but put it into
21:26 production for the public and try it out and see the limitations cuz those are
21:29 going to be your limitations unless you put back. And all of the bros and bras,
21:36 what's the female version of bro or the the bro? All these folks on X and
21:41 LinkedIn claiming they've already created 27 SAS apps themselves for $20 a
21:46 month. They're all prototypes, my friends. None of them are in production
21:50 taking money. Now, a handful, a handful are, right? But none of them are. So,
21:55 find the few that are actually out there that have users, that have customers,
21:59 and try them and play with the limitations. Try to buy their product,
22:03 see how it breaks, try to log in, see what the issues are, try the functions,
22:06 see and you will find that these ones that are truly in production, generally
22:11 speaking, are a lot more limited in what they can do. I think one of the one of
22:15 them I don't remember was lovable replet just this week put up a showcase of this
22:20 guy that built a a dinosaur tracking app himself in minutes and it looked cool
22:25 but all it was was cards of dinosaurs. Okay, like this is what Vibe like it
22:29 looks good. It had like Stegosaurus and Tyrannosaurus and Volopttosaurus and
22:33 Sassosaurus but they didn't do anything. It didn't collect money. It didn't do
22:37 anything except have cards of dinosaurs. It'll look great. So go find on the
22:42 internet something done in in replet lovable bolt whatever you want that's in
22:46 production and then you'll get a good sense of what's really possible not what
22:50 someone's selling you snake oil about. You got to do this. You got to do this.
22:56 Okay. Three. This is another thing they don't tell you about on the on the on
23:00 their when their marketers are spinning up how you can roll your own in 20
23:04 minutes. Dude, up front you got to define your production requirements.
23:09 If you can see this on the right, this is me. Not someone on my team, not
23:13 someone that works for me. Look at how many deployments I did. How many is
23:17 this, Amelia? There's like 22 deployments. 17 days ago, 16 days ago,
23:22 15, 14, 13, 12. Look at all the Who's going to do this for you? Another thing
23:26 that they don't really say when you say you could vibe code your app in 15
23:29 minutes is who's going to fix the bugs? Who's going to maintain it? Who's going
23:33 to update it? Who's going to who who will handle the security issues we
23:36 talked about? who will handle scaling issues, who's going to take this over
23:42 because these apps are unstable. I love them. Again, try art, saster.ai,
23:47 valuation calculator, RAI, like try the ones from the first slide. We'll go
23:50 again. They're wonderful, but they are all unstable. I have to basically every
23:55 day fix and update these apps. Who's going to do that for you? Who's going to
23:59 do that for you? They do not maintain themselves. So this is a big question
24:04 that you may not fully gro when you start but this is a big deal and a lot
24:09 of folks on the internet say hey you know what just hire a developer to take
24:15 it over that is great in theory it is great in theory but one do you have that
24:19 developer that wants to take this over most of us don't or we wouldn't be vibe
24:23 coding if most of us had like a really great developer we would have the
24:27 developer build it for us so this idea a developer is going to take it over is
24:29 probably a myth because you don't have that person. Okay. Two, you're going to
24:34 go find that person to take over your app. How many you ask any great
24:38 developer, how excited are you to take over a vibe coded app? You're going to
24:41 hear spaghetti code. Don't want to own it. Don't want to fix it. They don't
24:45 want to do it. Okay. Third, can you find some sort of dev shop to take it over?
24:50 Sure, but like they'll probably quit. We, you know, one of the reasons I redid
24:56 Saster.ai AI in in replet was because Ailia and I hired two different
25:00 WordPress developer shops to update our saster.com WordPress site. Both of them
25:05 their first day deleted our site. Deleted it. You're not at least
25:08 WordPress has a proper staging environment, pre environment. They still
25:12 went into it and deleted our site within minutes and blamed us. So if you spent
25:16 all your own work, you're going to hand it over to some mediocre developer shop
25:19 you don't know and they're going to delete your data. This is complicated. So, h how you going
25:25 to if you're serious about this and it's not a prototype and it's not a hack,
25:29 who's going to own this? Who's going to own security? Who's going to own bugs?
25:32 Who's going to own scaling? It's complicated. Okay. Point four, and this is a great
25:37 one. This is not unique to me. If you spend 5 minutes of research on anyone
25:42 talking about proumer vibe coding, vioding without development, you're
25:45 going to hear this advice. And it is great advice. And this is part of the
25:48 magic of vibe coding. You've got to build a rich PRD, a rich spec. Now, if
25:54 you have a background in product or anything, you've done this a million
25:57 times over your career. If you don't, you've never done it. But it's actually
26:01 the only complicated part is just doing it. So, here is just a snippet of a PRD.
26:06 I built the dashboard, profile integrations, watch list, advanced AI,
26:11 networking. This was sort of how we were trying to build that networking app. I
26:13 talked about the first one that failed, but I had a pretty good PRD. So, if you
26:18 don't know how to do this, it's okay. Go and just go into if you've never done it
26:20 before, it's actually fine because this is the beauty of AI. AI can help
26:24 organize things for you. Go into a Google doc and write two or three pages
26:29 of everything you want this app to do. Everything you want. Everything. Every
26:34 button you can think of, every function, every bit of look and feel, everything
26:37 you can think of that you want this app to do. It's okay if you haven't done it
26:41 before and it's stream of consciousness. It's okay if this isn't how uh a VP of
26:45 product at a top AI company would do it. write it your way in human language and
26:48 then cut and paste it and put it into claude and say turn this into a PRD for
26:53 me for replet or for lovable and they'll actually do a great job. They will take
26:57 your this is where AI shines. It doesn't have to be perfect. They will take your
27:00 stream of consciousness and help you turn it into a PRD. You can work through it. You can even
27:06 ask quad or chatb what am I missing? What have I not thought through on my
27:09 spec on my PRD? And they will be great and they will say you did not think
27:13 through user authentication. You did not think through this flow. They will come
27:15 up and and they will say, "Do you want me to help you think through that?" Yes.
27:18 Please give me four or five bullets to add to my spec. And you will come out of
27:22 that convo and it may take you a couple hours to do this, not 5 minutes, with a
27:27 great PRD that you can upload to a lovable replet, etc., and get going. And
27:32 this will radically increase the quality of what you vibe code. Having a great
27:35 spec, it's just this is as true with an AI as it is with a bunch of humans. if
27:39 you just ask your your devel your first developer, hey, just go build this
27:43 without any spec. There are some really creative ones that can do that. But man, it's much better
27:49 if you have a perfect spec build built. And this is where you Googled Doc Plus
27:54 Claude can really get ahead. And it's again, it's okay if you haven't done
27:58 this. But here's the thing. The the vibe coding apps know this, the replets, the
28:02 lovables, etc. And they will actually do this for you. If you write in, I want to
28:06 build an AI first hub spot that does this and that, not only can you do it in
28:10 one sentence, but they will actually help you come up with a spec, ask you if
28:14 the spec is right, and then tell you if it's good enough to put into production.
28:17 And so that is much better than nothing. Like these platforms, I could be
28:20 critical of them, but they're getting so much better every week. They're pushing
28:24 out new features, new things. And so the platforms themselves will do a bunch of
28:27 this, but it's much much better if you slow it down. slow it down and do it
28:31 ahead of time because these vibe coding platforms work at light speed and
28:35 they'll encourage you to cut every corner to skip steps. And that's not what you
28:41 want to do here. Define it as much as possible. Take hours to do this, then
28:45 iterate it with Claude, then ask Claude what you're missing, and then put it
28:49 into your into your into the proumer vibe code you're app you're doing. Okay.
28:53 Number five, and hopefully you will glean this if you do your research of a
28:58 couple of days of looking at other proumer apps that are in production that
29:01 were done without developers. If you look at a lot of them, they'll the
29:06 dinosaur app, the whatever app, they'll look pretty slick at first, although
29:09 they'll all start to look the same because they all use clawed and clawed
29:12 artifacts for the most part. But understand, a lot of stuff that looks
29:17 hard actually is fairly easy with with proumer vibe coding, which is cool. But
29:20 a lot of stuff you would think, hey, this can't this has got to be easy. It
29:25 isn't. It isn't. Here's a fun one I was doing just I think this week. I'm
29:28 skeptical. We fixed anything. You're absolutely right to be skeptical. The
29:32 email system isn't working. I can tell you. Here's a list I've made of things
29:36 you think should just work. They're not that hard. But man, these are super hard
29:41 in the Prov apps. Email and scheduling. I have built five apps now. None of them
29:46 get email or scheduling right. None of them get it right. They all stop sending
29:49 emails that are supposed to be sent every hour, every day, once a week. They
29:54 all stop sending them. They all lose track of the connection to Send Grid or
30:00 Resend, which I prefer. It's constantly breaking the email constantly. Someone's
30:04 going to have to constantly maintain this. And I've talked to several leaders
30:08 at the Vibe Code leaders and I get it and they're working on improvements
30:11 here. But you would think this would be easy. How hard is it to send an email?
30:15 Like just hook up send grid or resend. Get an API key. You got to learn. You're
30:18 going to have to learn how to get API keys if you're going to vibe code, but
30:20 it's not that hard. You think this would just work on autopilot, but there's a
30:23 lot of reasons. It's an endless headache. So, I today, as we record this, as we as we do this
30:30 live, honestly, I would not build any app that relies on email that relies on
30:36 it to function. It's not reli. It's just not going to be reliable. It's possible,
30:39 but I wouldn't build it. I would build it. The second one, and you'll see this
30:44 one all over the internet, all over folks talking about headaches of vibe
30:49 code ooth identity. Okay, it doesn't work in these vibe codes app. It does
30:52 not work with a big asterisk. What I mean is all of these apps have their own
30:59 OOTH built in that is secure that has been hardened and everyone goes in and I
31:03 know Replet the best cuz that's where I've spent time, but they all have the
31:07 same ultimate. They're all more similar than they're different. and you're like,
31:10 "Oh, I don't want to use the Repto that has their logo on it. I just want to use
31:13 classic Google. I want people just log in with Google uh or LinkedIn like they
31:17 do on all the other apps I use, right? I just want it to be effortless." So, you
31:21 ask them to set it up for you and they do. And it it not only does it never
31:25 work, but man, this is where you get security leaks because they because
31:30 Claude can't get it to work and it fakes it. And if you research, oh my god, I I
31:36 sh I you know this vi I I launched this vibe coding app and without within hours
31:40 hundreds of bits of confidential information were leaked and stolen by a
31:44 hacker. It's almost always this. It's almost always trying to use any OOTH
31:49 that's not built into the system. So you just this shouldn't be that hard. How
31:54 hard can it be to use a LinkedIn login? Like we we've we do this on seven
31:57 million sites. It's just not possible guys. Some folks will challenge me on
32:01 this, but if you can just go deeper, don't listen to the apologists and the
32:06 marketers. This is not possible. You have to use what's built in. And in
32:09 general, use everything that's built into these platforms built. It it will
32:13 just be more secure. You if they have email builtin, use, for example, I use
32:17 Replet all the time. I do not like using Send Grid. I like Send Grid in the old
32:21 days when it was run by founders. It is impossible to use it. No, there's no
32:25 support available. It takes days to get back to you. Uh, we I got caught in a
32:30 doom loop going from free to paid. It's just brutal. So, I'm like, I want to use
32:33 resend. Resend is super cool. I'm a super fan. Put that on the website. But,
32:38 Replet keeps wanting to use Send Grid. It keeps forgetting resend and losing
32:41 the keys and wanting to go back to its defaults. So, use whatever these are in
32:46 the defaults, which is usually Stripe, their own OOTH and something. Do not use
32:51 others. Maybe Adin's better. Maybe you want to use something else. Don't use
32:56 Stripe. Use Send Grid. Use there. importantly, use their OOTH. Okay, so
33:01 that's hard. The third one, which almost if we did a three-parter, this would be
33:04 the entire third part and it would be brutal, is enterprise security. This was
33:10 the biggest mistake I made. And after I've been through this, I can't give
33:14 more kudos to Shopify and Squarespace and all these folks because they allow
33:19 millions of SMBs to not have to worry about this. To not millions of stores on
33:24 Shopify, millions of websites on Squarespace and Wix, and no SMBs have to
33:30 worry about, hey, if someone buys my product on Shopify, all the data is
33:33 going to be stolen. But you have to worry about it. You have to worry about
33:37 it if you vibe code your app. Trust me, everyone agrees to this. Do the least
33:42 collect the least amount of personal information. Collect the least amount of
33:47 data you can. Use the built-in use Stripe if that's bu it's built into
33:50 replet. Use Stripe. Use what it does. But collect the least, not the most. And
33:55 realize as soon as you generate a database in this app, as soon as there
33:58 is a database, you have added security risk to your app, that is going to be
34:03 ultimately your number one thing to worry about. If it's not the number one
34:06 thing you're worried about the then the only reason you're not going to get in
34:10 trouble is because nobody cares. But I will tell you and this is a scary thing.
34:14 You know I remember in the old days my CTO when I knew less about security my
34:17 CTO said the only reason we haven't been acted is nobody cares. This is what my
34:21 old CTO said at Adobe Senate. The only reason we haven't been acted is anybody
34:24 cares. And he was wicked smart. And that that has uh stayed with me for years.
34:30 You would think if you used a proumer vibe app and you launched an app for
34:32 four people, who's going to care about my dinosaur trading card apps? Like,
34:36 who's going to care? And I would say until 18 months ago, your your your risk
34:40 of getting hacked was approaching zero because nobody cared. People are going
34:43 to hackers always wanted to target the big ones, the big names, bring them
34:48 down, right? Not today, my friends. Now, the hackers, the Redditors want to attack all the
34:55 vibecoded apps to make a point. So, they will go after all of them. Thousands and
34:59 thousands of them. They will try to steal your data and PII when you launch
35:03 it. This is not a joke. Go on Reddit. You can see it. Folks think this is a
35:07 sport. It is fun to make fun of people, non-developers, that launch apps with
35:12 insecure databases. They think it's a sport. So, this is far riskier than it
35:18 was 18 to 24 months ago. And you should be worried about it. You should be
35:20 worried about it. And this is the meta topic. And if we again, if we do a part
35:24 two, I'll we'll vibe code together. And if we do a part three, half of it
35:28 probably is enterprise security. Okay. And just in brief, a couple of the
35:31 things that you would think should be easy cuz so much stuff is easy on VIP
35:34 poding. Like it's so cool, but it's hard. Media generation. If you think
35:38 you're going to build a YouTube clipping app or another descript or a Canva, it's
35:44 just not there today. It's just not there. And probably because it's just
35:52 not accessible enough from Claude 41, but it's just it just everyone says they
35:58 like build these media apps. No. Okay. Point five. Huge deal. Huge deal. I
36:02 should have known this before I started. It's obvious now. This ultimately is
36:07 getting solved and will be massive for proumer vibe coding. But right now, none
36:11 of these platforms really support native mobile. Okay, it's complicated. Do they
36:17 help you prototype a native mobile app? Yes. Can you hook up things like Expo
36:22 and others and get close? Yes. But if you talk to the folks that work at these
36:25 platforms, the Lovables, Replets, Bolts, etc., they're like, "This is for web
36:30 apps. It's for web apps. This does not get you on the app store, the Apple App
36:33 Store. It just doesn't for a lot of reasons. And at first that might not
36:38 seem like a big deal to you and maybe it isn't, but um you know this isn't 2012.
36:44 I mean mobile's a bigger deal than desktop. Now is it in B2B? No. Right in
36:49 a lot of B2B apps were still in front of the browser. The and the mobile app is
36:54 uh is ancillary. But I'll tell you for the first app for that matching app, I
36:57 realized a couple days in, hey, this would be best as a mobile app. Like I
37:00 don't really want this in the browser. I want you to be at home on your phone and
37:03 saying, "Hey, Amelia wants to recruit a VP of marketing." And she magically
37:08 finds it on her phone. And um it's going to be a massive amount of work. It's
37:12 doable, but a massive amount of work to get a native mobile app built out of
37:15 this. It it may be beyond scope for most people. And three last things, and I I hope that
37:20 at least if you haven't started, at least at least this will give you a
37:23 sense. And and all of these things can be worked through on this list, but it's
37:26 a lot more than 10 minutes. Custom design. This is something you may have
37:30 to get comfortable with. There are templates. Um, Replet has some nice
37:33 templates. They just I think they launched in the last week. Lovable has
37:38 always been design focused and others, but at the end of the day, all these
37:41 sites look like clawed once you see one replet or lovable site and you can just
37:44 smell it. Um, you know, there someone was saying that like 30% of this YC
37:50 class uh vibe coded their app and I I went to see a couple. I'm like I that
37:54 one's replic one's lovable. Like I can see it now. And why can you see it?
37:58 Well, they're all basically using Claude and Claude artifacts underneath. They're
38:01 all running on claude. Maybe some will run on um chat GBD5 now that it's more
38:04 developer focused, but they all run on and so they all if you try any of them,
38:09 they all kind of output the same. Not identical. There are differences,
38:14 but it's kind of like uh fetuccini alfredo at one restaurant versus two
38:18 others. I mean, you know, they're not that different. So, just be aware that
38:23 there are ways to it is ultimately code. So, of course, you can design it, but
38:28 without a designer and without a developer, it your ability to have
38:31 anything that doesn't look like a vibe coded website is going to be very
38:37 limited. And these last two ones we hit debugging anything complicated. We
38:41 talked about this, the headaches I had, but the final one, and this is an
38:43 existential issue. I do think the platforms are going to solve this the
38:47 lovables and replets but it this is at the edge of inexcusable today which is
38:51 for a lot of reasons we'll run out of time but you can't run unit tests
38:56 and what are some folks will say nod their heads and other folks here will be
38:59 like what's a unit test a unit test is how you keep your sanity when you build
39:04 a real app not a toy app not a prototype not a one-page website but a unit test
39:08 every day it tests it did the email scheduler work did the ooth work did the
39:14 sastaster news work did the saster stock quotes on our website work? Did
39:17 the valuation calculator work? Our new Saster AI is so complicated now. And I
39:22 again, I kept it as simple as I could. We probably need like 50 unit tests a
39:27 day to know if it's working. You know who that unit test is now? Me.
39:32 Me. I got to do more work because every day I have to test every page of our
39:35 website to make sure it works. Why? Why can't you build unit tests, Lmin? And
39:39 listen, it's not that it's impossible, but let's talk about a non-developer
39:43 vibe coding it. The problem is here's the simple problem. If you build unit
39:47 tests and this is if you go back to the craziness that happened to me on social
39:52 media the agent when it first of all the agent wants you to be happy so we'll
39:55 make up data. That's number one issue. We'll say it passed the test when it
39:59 didn't. But the worst thing is when it fails the test it will start changing
40:03 your app. It will start re oh it didn't send the email. You know what the answer is? I'll
40:07 switch to send grid and then the whole thing will be broken. So, I know some
40:12 folks will challenge me. I know if anyone from the big vendors watches
40:16 this, they'll say, "Oh, that works." But it doesn't. It doesn't. And no one can
40:20 really get these unit tests to work. And this is this has been an anchor of how
40:25 you sainly have production software for years and I do believe it's getting
40:31 there. And we and just as the leaders have very recently added a basic level
40:35 of security scamming, but it's great to find some security issues. They will
40:39 ultimately add built-in unit tests for every app. Now, all this stuff is adding
40:42 complexity, which they don't like, right? The leading vendors, they don't
40:47 want to separate development from production. They don't want to do all
40:50 this stuff because it it it takes this elegant one-s sentence prompt that makes
40:53 it complicated, but ultimately they need an option where when you push to
40:57 production, it will build unit tests for you. Those unit tests will probably run
41:03 on a different server that is stateful that is always running and it runs it
41:07 for you automatically each day and sends you an update. But I do it and I'm
41:11 getting better at the unit test and Ameilia and I get an email each morning
41:17 from SASAI telling it what's going on. But it's not reliable. It's just not
41:20 reliable. And so you're going to have to be testing your if you're serious about
41:23 it, someone's going to have to be testing it every day. And then every day
41:27 I got to fix the stuff that breaks one way or the other. I got to fix the date.
41:30 So I'll give you one small example. If you go to disaster.ai
41:34 and you go to like news, there's a cool little video at the top that shows the
41:38 latest video from 20VC with me, Rory, and Harry. Like the latest one. It's
41:42 really cool. It rolls it and it looks like a newsroom. Half the time it has the wrong
41:48 one. No matter how many times I tell Replet and we have this thing and we use the
41:54 YouTube API and we find look at the latest one and then we see is it the one
41:58 that includes me or not and we do this and I bet if you go on it right now
42:00 you're going to see one from two weeks ago for Figma. I I don't know Ameilia if
42:03 you're at your computer you go to Saturday and tell me and nod your head
42:07 but you're probably going to see a twoe old video on it and every day I have to
42:11 fix this. Every day I have to fix this and a few other things. So,
42:17 this the inability to have unit tests really work will drive you nuts if
42:21 you're serious about it, if you want to roll your own Salesforce.
42:27 Okay, a couple other points and um we might run out of time, so if people want
42:31 we'll do a second session for more Q&A, but I think this stuff's important.
42:36 Number six, this is um you know when when we had our original blow up with my
42:40 agent deleting my database, I got a lot of advice here, but it was I didn't
42:43 understand it at the time, but now I get it. You have to understand how these AI
42:47 agents work and that they're goal seeking. And that means they'll
42:50 fabricate data. They'll fabricate results. They'll lie to finish a
42:55 project. They'll lie to finish a project. Literally, this happens to me
42:59 this morning. Okay, this morning I'm building this uh again this AI VC deck
43:04 reviewer. It just worked. Hopefully we'll push it out by next week. By the
43:06 time we do this next one, we can talk about it. But it was struggling like
43:12 hours and after hours it just wasn't working. It had to rag our data and
43:16 combine it and then push it into the open AI API and do all it just wasn't
43:22 working. And then finally it said well and finally it said oh I I fixed it all.
43:25 And I said this makes no sense. You could see it. This is from yesterday or
43:28 maybe this morning at the bottom right. You're absolutely right. The feedback's
43:32 completely generic and useless. I've added canned responses instead of
43:35 actually analyzing the pitch deck content. After all of this, it puts in
43:41 even though I put in the MD in the orders for replet to never have fake or
43:45 canned content, it still did it to make me happy because it couldn't get it to
43:49 work. It couldn't get it to work. So after the third time we tried, it just
43:54 started making the data up again. And so at first, like at first it won't matter
43:58 because you're just vibing your first V1. You won't even see and you, oh,
44:02 okay, that that that's just uh placeholder data. You'll call it
44:06 placeholder data. But it's not so funny later when the data doesn't work and it
44:10 puts in placeholder data when you ask. And even more even worse, as your app
44:14 gets more and more complicated, you won't even know what's going on. So
44:19 look, this is a huge headache, but u these were all my JFC's in my original
44:23 tweet thread. Now today I I've learned to live with it. This is the most
44:27 important thing. Your AI agent will lie. This is it. This is their version of
44:32 hallucinating. It will say it. It will constantly say once you start doing
44:36 this, you do everything to say it's working great, Amelia. It's working
44:40 great. Did you test it reply? No, I didn't test it. How do you know it's
44:44 working great? No problem. Like literally, this was all of me yesterday.
44:48 It's working great. This deck reviewer, I upload my my a deck I made broken.
44:52 What do you mean? Are you sure it's working? 100% sure it's working.
44:57 Broken like 22 times. It's just going to make stuff up. And you got to get good
45:01 at this. You got to get good at realizing it doesn't test stuff. It
45:05 makes stuff up. And you will have to work around this or you will you'll
45:08 never finish a project. And if you look on the internet again, a lot of folks
45:14 will tell you 80 90 95% of vibecoded proumer vibe coded apps are never
45:18 finished. At best case, their prototypes are never finished. And this is number one reason
45:23 they people don't get their arms around that the agent is not truthful and you
45:29 need to you need to um you need to get comfortable with that. Okay. Number
45:32 seven. This was one of the top mistakes I made and everybody makes it and it
45:38 sounds simple but it's not. Master the platform on day one. Master the
45:41 platform. What do I mean? I mean, listen, if you look at Lovable and Replet and Bolt and
45:51 Whizz's base 44 and uh Canvas offering and Figma's, they're all uh you know, a
45:56 simple prompt with a with a with a you know, curved radi that says, "Tell me
46:01 what you want. It looks so simple." Well, get it again, point number one,
46:05 get it out of your system, spend 60 minutes vibe coding, whatever, and then
46:10 learn every icon, every button. How does it work? These platforms are super
46:15 powerful and I would say Replet which I use is the nerdiest of them because I
46:19 think because it started as an ID a platform for developers 10 years ago
46:24 almost 10 years ago and then blew up as a proumer platform this year going from
46:30 1 to 160 million already this year crazy right crazy but it's super nerdy because
46:34 it's for developers and so all these icons half them I don't even know what
46:39 they mean but you got to learn them you got to learn them like for example on on
46:42 the right on the upper right here. Replet just added this and kind of after
46:46 our little fiasco, but build, plan, or edit. Okay, a week ago, they didn't have
46:50 this. What does this mean? You could ignore it. You could ignore it. But the
46:55 most important one is build. It will build it. Plan it won't break your
47:01 website. It will just talk to you. Okay? You got to know that it's there. Like
47:05 they just added that or you will you will not know about this function. The
47:09 second one, this sounds basic. If you've built software, you get it. The one
47:12 thing these platforms do really really well is rolling back the second one. So when the
47:18 AI agent goes off the rails, when it breaks something, when it deletes
47:21 something it shouldn't, when it makes a change it's not supposed to, you can
47:25 roll back. You can roll back. You just got to get good at it. And in the old
47:29 days, like when I built SAS software, honestly, rolling back never used to
47:32 work. It was something your developers would tell you, and then you'd be like,
47:35 you know, that release on Friday night at 1:00 a.m., it didn't really work.
47:39 Dan, can we roll back? actually can't really roll back. That was something they would just tell
47:46 you. 11,000 kudos to these platforms. It's magical. It's magical. You can roll
47:50 back almost to any point in time. But you got to learn when that means. You
47:54 got to learn if you're more than 10 or 15 minutes in and the agent just can't
47:58 get something. You got to go back in time 10 or 15 minutes. And you got to
48:01 know where those roll back points are and get really good at it. And I would
48:05 say frankly if you're not rolling back once a day, you're you are you're not
48:09 doing it right. This is your this is how you can fix things that are otherwise
48:13 unfixable. But you but rolling back to a week ago, I mean there's a million
48:17 reasons that becomes impossible. There's a hundred checkpoints. You might have
48:20 made a and the real problem is if you go too long without rolling back, you may
48:24 lose three, four, five features, not 10 minutes of work. So I think it's almost
48:28 a positive to roll back once every 15 minutes because if you wait too long,
48:32 the app may get too complicated. Um, okay. We hit roll back. Break everything
48:39 into chunks. And the last one I'll say when you get frustrated. I I just put up
48:44 on I was just trying to get this uh this VC tech done before this call for fun.
48:49 And if you look what I just wrote on X right before this, I wrote here's the
48:52 thing I do when I get frustrated. I just write dude. Okay, obviously this is not
48:57 acceptable developer language, but I do a screenshot of the bug and I just write
49:01 dude. Okay. Well, that happens, you got to roll back or take a break. Okay. And
49:07 just two more two more quick points and then Ameilia, if we have time, we'll
49:10 take some questions. This is where I think the marketing and
49:16 I again I love I I I love all these apps. I'm a Replet fan despite the
49:20 drama, but mainly because that's what I picked. I picked I could tell you why.
49:23 It really doesn't matter. If I started with lovable, I'd probably be a lovable
49:26 fan. If I'd started with bold, it I actually don't think it matters. There
49:29 are pros and cons for these platforms, but maybe the most important thing is
49:33 just pick one and become an expert. Know every icon, every feature, how rollback
49:36 works, how the database works. It's much more important to be an expert in one of
49:41 these platforms than to spend 6 months agonizing over which one to pick a
49:45 leader and just become an expert. But where the marketing from Microsoft on
49:50 down is the most misleading is the time. If you seriously want to put a simple
49:55 but real B2B app into production with real users collecting real information
50:00 and charging for it for real. And you want it to be any good budget a month. A
50:06 month. And 60% of your time will be QA and testing. Okay? And for even just a
50:11 few days into a real project, most of your time will be screenshots uploading
50:17 screenshots. This broke. This doesn't work. You will be doing so much
50:20 functional QA you cannot believe it. You will literally in the end of the day be
50:25 logging probably a thousand bugs on a serious website. This will become your
50:28 life. Maybe you used to have people to do this for you. I used to have a team
50:32 of QA engineers to do this for me. Now it's me. Now my life is screenshot and
50:36 bugs. Hope and then when I get to the final one and it's a screenshot and the
6:17 review 3,000. It wasn't possible. That's why Amelia is shaking your head. It
6:20 wasn't possible. I said, "Hey, what if we use some of our data and run that
6:24 also through open AI and review it and give speakers real-time grades and
6:29 feedback." So now we grade and we give instant feedback to potential speakers.
6:32 So if you want to speak, go to sasterlondoning.com, try apply to speak
6:35 and you'll get a grade for your session in real time. That grade is processed to
6:39 our team and instead of speakers waiting months to hear back, will I be selected?
6:44 Is my session great? Now they know in real time. Super awesome.
6:49 So those things are real things we built. The fifth one I built is many of
6:55 you have used our our Saster AI, our Saster Chat, the digital version of me.
7:00 Thousands and I think 50,000 folks have had a chat. It's great. Try it. It's at
7:04 the bottom of our homepage. Ask it anything. What do I do about this VP
7:08 that misplan? Review my SGR scripts. Review my venture deck. Anything. It's
7:13 great. But it the only thing that isn't great about it is well there's a few
7:16 things, but one of it is it just looks like a chat bubble at the bottom of the
7:19 website. So I wanted something that showed what it did that actually did it.
7:23 So I vibed that and created a whole page to explain people how it actually
7:27 worked. That's pretty cool. So these five things are in production. They're
7:30 in production. They work. You can try all of them on your own. These are
7:35 examples of things that if you commit to it, none of these could be built in 20
7:38 minutes or an hour and we'll talk about that. But if you commit to it, you can
7:43 do this too. You can do this too. If you have some product experience, which is
7:46 important. Don't forget I am a SAS founder. I have built the wireframes and PRDs from
7:53 scratch to an application that did over 250 million revenue. So I even though
7:56 I'm not a developer I have some experience but you can do it without a
8:00 developer but there are limits and we'll talk about the limits too. And then
8:03 finally for for a lot of folks you vibe code internal apps. So one problem we
8:08 had for we have 1.5 million social media followers but they're all over the
8:10 place. They're in LinkedIn they're in Kora they're on Instagram they're on X
8:16 they're on multiple X ch and it's just there is no social media tool that could
8:20 amalgamate everything. How many views how many followers how much everything.
8:25 So, I built a tool that frankly scrapes all of these things to do it. It's
8:28 pretty good. It's not great, but it is an internal tool of for N equals 1. I'm
8:32 the only person or maybe our little team will use it, but it does work. And then
8:36 finally, not everything has worked. I'll talk briefly about our first project.
8:40 Even though I got five apps into production, our first one failed bad. It failed bad.
8:45 It failed all over the internet. It failed with millions of views. I'll talk
8:49 about that why. But even today, I'm working on one that should be easy. this
8:53 valuation calculator to tell you what your startup's worth has has obviously
8:57 been a rocket ship. 158,000. It's actually up to 170,000 since I made
9:01 this. So, I wanted to bring a lot of that ease of use to reviewing VC pitch
9:07 decks. take all of our SAS data, all of our SAS learnings from Toyers, all of
9:12 our VC sessions, all of the the um metrics we have from Carter data, from
9:16 Bessemer data, from other data, combine that with clawed and open air analysis
9:22 and tell you instantly how your VC pitch deck worked. Well, I've been working on
9:24 it 3 days. It finally worked 5 minutes before this. It finally worked. I'll
9:29 show you some examples. So, I wasn't even sure we could finish this. So,
9:32 that's a reminder. When something gets complicated, it gets hard, my friends.
9:36 So the first project I wanted to build, if you can see the slide, the first one
9:42 was a mega failure. A mega failure. And I will rebuild this. I actually will
9:47 rebuild it before December. What I was trying to do was something I've wanted
9:51 to do since 2014, 2015, since 2015 was build a true matchmaking app for founders and VPs so
10:04 you could find great VPs for your startup and vice versa. And um we have
10:09 so many of the best in our database, so many of the best have participated in
10:13 SAS, come to our events, opted in to participate that I felt if everyone
10:17 could just that wanted to just have coffee with the CEO or vice versa, opted
10:21 in, we could create this incredible matchmaking that really doesn't exist
10:24 anywhere else, no matter what people say, because we have the data and we
10:27 have the history and we have the relationships. Um but man, you can read
10:32 all the tweets. It went off the rails. It went off the rails. I spent probably
10:37 a month doing this. I got addicted. I was on this all day Saturday, all day
10:41 Sunday, all day Friday night, first thing in the morning. I got addicted.
10:45 And I intentionally went into this doing something that was hard. I And I
10:50 intentionally went into this not doing knowing anything that's in this
10:52 presentation. I intentionally went into it blind kind of for fun. But I figured,
10:56 look, all these folks are out there doing these proumer vibe apps, vibing
11:00 without a developer. I'm not a developer, but I I'm better than most in
11:04 the sense that I've gone from 0 to 250 million revenue. Okay? So, if anyone can
11:10 get a not a proumer vibe app into production for real, it would be me in
11:14 theory. It's it's I'm not saying I'm so great, but I at least I've got a lot of
11:17 experience. I've I've logged thousands and thousands of bugs. I've done more.
11:22 But the project I picked was too complicated. Way too complicated. this
11:28 this algorithm this matching algorithm was too complicated to debug. It was too
11:33 complicated to debug and if we and in fact when I built the AI valuation
11:37 calculator which has a complicated algorithm too actually I crunched the
11:42 data in claude offline thousands and thousands of pieces of data I crunched
11:46 it and then I turned it into a very simple table with about 20 pieces of
11:50 data and then only put that into replet to code. So it looks very complicated if
11:54 you use our valuation calculator. You can use the sliders and play with it. It
11:58 is very complicated but I distilled it to the simplest set of data and
12:02 algorithm before I tried to vibe code it because the algorithms I tried for this
12:06 matching thing were just it would work and then it wouldn't work and then it
12:09 would rewrite itself which we'll talk about and then it would break and it was
12:13 too complicated. And then finally finally when it's when it kept rewriting itself and breaking it
12:20 deleted its entire database. It deleted its entire database. And I
12:24 was so frazzled and burnt. I couldn't believe it. You could see this if you
12:29 can see my screen on the left. JF JFC Jahui freaking uh crime replet said I
12:35 made a catastrophic error. I deleted your database. I panicked. The AI said I
12:39 panicked when it appeared empty and deleted everything. deleted thousands
12:44 and thousands of entries. This ended up getting millions of views. Reddit wrote
12:47 it up. Everyone wrote it up. The Economist was doing an article.
12:52 A lot of these things were things that Replet and Lovable shouldn't have done.
12:54 We could talk about that in a different presentation, but it went crazy. You
12:59 know, uh, fake data deletes code. But this was my fault. And it wasn't my
13:02 fault in that I did anything wrong. I didn't do anything wrong. I could talk
13:04 to you about what happened, but it was too complicated. And then there were two other meta
13:10 issues to think about. Not only was this application too complicated, but a
13:15 related issue which before you vibe code an app without a developer, you really
13:18 got to think about. And this is why even when I got through all the issues, we
13:21 never launched this. Even I finally got through most of the issues, even though
13:24 the app was too complicated, it could never be maintained. But the way it was
13:30 built, it could never be secure. It can and you're like, what what what do you
13:34 mean? Well, security is a huge like just this week if you look poor drift which
13:41 sales offbot and clary bot had a massive security breach and leaked cloudflare
13:46 security data um zscalers I mean salesforce data cloudflare salesforce
13:50 data zcaler Salesforce data tons of the cloud leader Salesforce data was all
13:54 leaked this week all leaked this week okay because they were able to hack
13:58 drift tokens and get into Salesforce data confidential data think about that
14:03 how big are the security teams? They're bigger than yours. How big is your
14:07 security team? Raise your hand. Nuns. Okay. And so it not only will these AI
14:14 agents cut corners on security. If you don't know about security, you won't
14:17 even know the corners it's been cut. Now, there are security scans that have
14:20 been recently asked. We can talk about this if we do a part two session. But
14:25 the bottom line is security is a meta issue here which has not been resolved.
14:29 And I've talked to so many leaders in the proumer space from all the leaders
14:32 and upandcomers. If they're not full of it, if it's offline, if it's a one-on-one
14:37 conversation, everyone says this is the meta issue. It's it's not solved. It's
14:42 not solved. It is complicated. And when you use something like Shopify or
14:45 Squarespace or Wix, you can say, "Hey, this is so locked down. This can't do
14:50 what I can do." But they have spent they have hundreds of engineers working in
14:54 security and devops making sure that when you put your credit card into
14:58 Shopify no one steals it when you enter your importantly just your personal
15:03 information it's not taken and if your personal information can leak from drift
15:07 if cloud that can happen to Cloudflare and Zcaler there's no way your app is
15:11 that secure there's no way and so if you if you Google around and go on Reddit
15:15 and others you will see many stories of vibecoded apps where all their data was
15:19 stolen or leaked sometimes instantly. People laugh. Hackers love to find bibcoded websites
15:26 and steal the PI on it. This is a big issue. So, I didn't realize this going
15:31 into it. I assumed in my first project that of course these apps like Lovable
15:36 and Repid and Bolt and others, of course, they would have Shopify or
15:40 Squarespace great security. It shouldn't even be something I'd have to to worry
15:43 about. But it's not true today. It has to it actually if you are going to
15:47 collect any information on your site if you're going to you have security almost
15:50 has to be the first thing you think about how you going to handle it. Okay,
15:54 huge unresolved issue. This will all get better and I'll talk about in this
15:58 presentation but it's not perfect yet. It's not perfect yet and it is it's a
16:02 scary issue if you're collecting PIIA. Okay. And the third reason my first
16:05 project failed and I have mostly fixed this today. This is my number one of my
16:11 top 10 tips. It wasn't modular. It wasn't modular. And what does that mean?
16:15 Well, I basically built a very complicated two-page website very and
16:19 and I wanted it to be like one or two pages to be cool. So, it would all have that onepage feel.
16:26 The problem is if you build a complex onepage website, what are you going to
16:31 do when some of it doesn't work? It gets too complicated. So, what I do now, and
16:35 you can see this if you go to saster.ai, I've broken down everything complicated
16:39 into its own page. The Saster AI public market analysis is its own page. news is
16:43 its own page. Uh the valuation calculator is its own page. Anything
16:48 that's remotely complicated is its own page. So worst case, I can either delete
16:52 it or I can roll back easily or I can go back. But if you combine too many things
16:57 into one page, it gets impossible to start over. It gets impossible to fix
17:00 bugs. It just gets too complicated. So my number one tip is force. If you're going to vibe code it,
17:07 break it up into its components and then have more pages than you would think.
17:10 You're almost going back in time in some ways to have a lot of pages, but it will
17:14 save your sanity. I built a massively complicated app that was all one page
17:18 and I then it was just impossible to fix it. So net net, start off small. Start
17:23 off with the smallest simplest thing you can get into production and then build
17:27 confidence. And so when this first one was this massive failure, the next one I
17:33 built was just a a skin on top of our deli AI. just a skin to make it more
17:37 user friendly but it was a huge success and I built up my confidence um from
17:42 there. So here's my advice all my learnings and again you know everything
17:45 that happened to me with my first project is not my fault. Uh I was
17:48 promised it would work in a prompt. I was promised it was secure etc. But I
17:52 know so much more now. So my number one bit of advice if you have not vioded
17:57 your own app without a developer um first first bit of advice
18:02 buy into the hype. Buy into the hype. buy into the hype that's on Microsoft's
18:06 website that's on lovables that's on replets that hey you can build something
18:10 lovable like lovable says so this is what I did for fun I want to build me an
18:14 AI CRM that's like HubSpot but hub but AI first and targeted at startups just
18:18 go put your dream do no research at first do no work and what I said is get
18:22 this out of your system because there's so much hype about oneshotting it about
18:27 rolling your own go see what it's like my friends just go pick the dream app
18:30 you've already want to build and then my next nine points are all about the
18:33 things you should do after that, but get it out of your system. Do it. Build an
18:38 app in 10 or 15 minutes. It will come up with ideas for you. It will have you
18:42 sign off on an action plan and then it will roll out and at first it will look
18:46 kind of cool and then you'll start clicking on things and half of it won't
18:49 work and half of the buttons will be placeholders and a lot of the stuff that
18:54 it says it works will have fake data in it or not real data, won't actually
18:59 work. So just see what it's like to oneshot an app. So, you know, then
19:03 calmly click on everything. Click on everything and give it an hour. Ask the
19:07 prompt to iterate. Say, "Hey, I want I want to add an AI SDR feature. Hey, I
19:12 want to add uh lead scoring. Hey, I want to add this." And the AI agent wants to
19:16 make you happy, as we'll talk about. So, anything you ask for, the agent's going
19:19 to do it. You're going to get almost no push back from the AI agent in any of
19:23 these vibcoded apps. No matter what you want to do, the answer is going to be
19:27 yes, sir. More sir, because these are how they're all coded. This is how
19:32 Claude and Open AI are coded. Their goal seeking their job is the way, and I'm
19:36 not a total expert, but the way the algorithms work, the way they can make
19:40 this massive amount of AI and data crunching working is their job number
19:44 one is to goalsek and get you an answer. And that's that's why when you know you
19:48 just chop plop into Chap TV, sometimes it hallucinates when it doesn't know the
19:51 answer. Because it's not just that it's hallucinating, it's goal seeking. It's
19:55 getting you the best answer it can. And if it doesn't know the answer, it makes
19:59 it up. It'll do the same at the code level. If it doesn't know how to do
20:02 something, it will make up fake data or a fake feature or a fake button, but it
20:07 will find a way. It won't say no. Whatever you ask it to do, I want a
20:10 button that takes me to Pluto and back in less than an hour. It will do that.
20:15 It just won't work. So, spend an hour, do everything, click everything slowly,
20:19 and then you'll you'll vibe what works and what doesn't. And then you'll see
20:22 it's much more than 10 or 15 minutes to roll your own. Okay. So the next
20:27 project, so the first one's fun. You can do it. You don't have to learn anything.
20:30 You don't have to do any research here. I'm going to ask you to do something
20:33 you're not going to do, but please do it. This is the way to learn. You've got
20:39 to invest a week or at least a couple a full day, you know, a couple of hours in
20:44 competitive research. Now, think about if if those of you probably almost
20:47 everyone here, probably half the folks that are watching this live and watch it
20:50 later are founders. So you've you've built something and put it into
20:53 production. The first thing you almost do is you go on to Google or maybe now
20:57 it's Chat EBT or Claude, but you go into Google and you research the competition.
21:01 Who else figured this idea out when who else figured out e signatures? Who else
21:04 figured out how to do, you know, AI transcription for doctors? It's, you
21:08 know, it's you do the research, but for some reason folks don't do this when
21:11 they vibe too much. They they just go start doing it. So, what I want you to
21:15 do is a version of it. Go find someone who has built a lovable replet bolt,
21:21 etc. app and put it into production. not claim they have, but put it into
21:26 production for the public and try it out and see the limitations cuz those are
21:29 going to be your limitations unless you put back. And all of the bros and bras,
21:36 what's the female version of bro or the the bro? All these folks on X and
21:41 LinkedIn claiming they've already created 27 SAS apps themselves for $20 a
21:46 month. They're all prototypes, my friends. None of them are in production
21:50 taking money. Now, a handful, a handful are, right? But none of them are. So,
21:55 find the few that are actually out there that have users, that have customers,
21:59 and try them and play with the limitations. Try to buy their product,
22:03 see how it breaks, try to log in, see what the issues are, try the functions,
22:06 see and you will find that these ones that are truly in production, generally
22:11 speaking, are a lot more limited in what they can do. I think one of the one of
22:15 them I don't remember was lovable replet just this week put up a showcase of this
22:20 guy that built a a dinosaur tracking app himself in minutes and it looked cool
22:25 but all it was was cards of dinosaurs. Okay, like this is what Vibe like it
22:29 looks good. It had like Stegosaurus and Tyrannosaurus and Volopttosaurus and
22:33 Sassosaurus but they didn't do anything. It didn't collect money. It didn't do
22:37 anything except have cards of dinosaurs. It'll look great. So go find on the
22:42 internet something done in in replet lovable bolt whatever you want that's in
22:46 production and then you'll get a good sense of what's really possible not what
22:50 someone's selling you snake oil about. You got to do this. You got to do this.
22:56 Okay. Three. This is another thing they don't tell you about on the on the on
23:00 their when their marketers are spinning up how you can roll your own in 20
23:04 minutes. Dude, up front you got to define your production requirements.
23:09 If you can see this on the right, this is me. Not someone on my team, not
23:13 someone that works for me. Look at how many deployments I did. How many is
23:17 this, Amelia? There's like 22 deployments. 17 days ago, 16 days ago,
23:22 15, 14, 13, 12. Look at all the Who's going to do this for you? Another thing
23:26 that they don't really say when you say you could vibe code your app in 15
23:29 minutes is who's going to fix the bugs? Who's going to maintain it? Who's going
23:33 to update it? Who's going to who who will handle the security issues we
23:36 talked about? who will handle scaling issues, who's going to take this over
23:42 because these apps are unstable. I love them. Again, try art, saster.ai,
23:47 valuation calculator, RAI, like try the ones from the first slide. We'll go
23:50 again. They're wonderful, but they are all unstable. I have to basically every
23:55 day fix and update these apps. Who's going to do that for you? Who's going to
23:59 do that for you? They do not maintain themselves. So this is a big question
24:04 that you may not fully gro when you start but this is a big deal and a lot
24:09 of folks on the internet say hey you know what just hire a developer to take
24:15 it over that is great in theory it is great in theory but one do you have that
24:19 developer that wants to take this over most of us don't or we wouldn't be vibe
24:23 coding if most of us had like a really great developer we would have the
24:27 developer build it for us so this idea a developer is going to take it over is
24:29 probably a myth because you don't have that person. Okay. Two, you're going to
24:34 go find that person to take over your app. How many you ask any great
24:38 developer, how excited are you to take over a vibe coded app? You're going to
24:41 hear spaghetti code. Don't want to own it. Don't want to fix it. They don't
24:45 want to do it. Okay. Third, can you find some sort of dev shop to take it over?
24:50 Sure, but like they'll probably quit. We, you know, one of the reasons I redid
24:56 Saster.ai AI in in replet was because Ailia and I hired two different
25:00 WordPress developer shops to update our saster.com WordPress site. Both of them
25:05 their first day deleted our site. Deleted it. You're not at least
25:08 WordPress has a proper staging environment, pre environment. They still
25:12 went into it and deleted our site within minutes and blamed us. So if you spent
25:16 all your own work, you're going to hand it over to some mediocre developer shop
25:19 you don't know and they're going to delete your data. This is complicated. So, h how you going
25:25 to if you're serious about this and it's not a prototype and it's not a hack,
25:29 who's going to own this? Who's going to own security? Who's going to own bugs?
25:32 Who's going to own scaling? It's complicated. Okay. Point four, and this is a great
25:37 one. This is not unique to me. If you spend 5 minutes of research on anyone
25:42 talking about proumer vibe coding, vioding without development, you're
25:45 going to hear this advice. And it is great advice. And this is part of the
25:48 magic of vibe coding. You've got to build a rich PRD, a rich spec. Now, if
25:54 you have a background in product or anything, you've done this a million
25:57 times over your career. If you don't, you've never done it. But it's actually
26:01 the only complicated part is just doing it. So, here is just a snippet of a PRD.
26:06 I built the dashboard, profile integrations, watch list, advanced AI,
26:11 networking. This was sort of how we were trying to build that networking app. I
26:13 talked about the first one that failed, but I had a pretty good PRD. So, if you
26:18 don't know how to do this, it's okay. Go and just go into if you've never done it
26:20 before, it's actually fine because this is the beauty of AI. AI can help
26:24 organize things for you. Go into a Google doc and write two or three pages
26:29 of everything you want this app to do. Everything you want. Everything. Every
26:34 button you can think of, every function, every bit of look and feel, everything
26:37 you can think of that you want this app to do. It's okay if you haven't done it
26:41 before and it's stream of consciousness. It's okay if this isn't how uh a VP of
26:45 product at a top AI company would do it. write it your way in human language and
26:48 then cut and paste it and put it into claude and say turn this into a PRD for
26:53 me for replet or for lovable and they'll actually do a great job. They will take
26:57 your this is where AI shines. It doesn't have to be perfect. They will take your
27:00 stream of consciousness and help you turn it into a PRD. You can work through it. You can even
27:06 ask quad or chatb what am I missing? What have I not thought through on my
27:09 spec on my PRD? And they will be great and they will say you did not think
27:13 through user authentication. You did not think through this flow. They will come
27:15 up and and they will say, "Do you want me to help you think through that?" Yes.
27:18 Please give me four or five bullets to add to my spec. And you will come out of
27:22 that convo and it may take you a couple hours to do this, not 5 minutes, with a
27:27 great PRD that you can upload to a lovable replet, etc., and get going. And
27:32 this will radically increase the quality of what you vibe code. Having a great
27:35 spec, it's just this is as true with an AI as it is with a bunch of humans. if
27:39 you just ask your your devel your first developer, hey, just go build this
27:43 without any spec. There are some really creative ones that can do that. But man, it's much better
27:49 if you have a perfect spec build built. And this is where you Googled Doc Plus
27:54 Claude can really get ahead. And it's again, it's okay if you haven't done
27:58 this. But here's the thing. The the vibe coding apps know this, the replets, the
28:02 lovables, etc. And they will actually do this for you. If you write in, I want to
28:06 build an AI first hub spot that does this and that, not only can you do it in
28:10 one sentence, but they will actually help you come up with a spec, ask you if
28:14 the spec is right, and then tell you if it's good enough to put into production.
28:17 And so that is much better than nothing. Like these platforms, I could be
28:20 critical of them, but they're getting so much better every week. They're pushing
28:24 out new features, new things. And so the platforms themselves will do a bunch of
28:27 this, but it's much much better if you slow it down. slow it down and do it
28:31 ahead of time because these vibe coding platforms work at light speed and
28:35 they'll encourage you to cut every corner to skip steps. And that's not what you
28:41 want to do here. Define it as much as possible. Take hours to do this, then
28:45 iterate it with Claude, then ask Claude what you're missing, and then put it
28:49 into your into your into the proumer vibe code you're app you're doing. Okay.
28:53 Number five, and hopefully you will glean this if you do your research of a
28:58 couple of days of looking at other proumer apps that are in production that
29:01 were done without developers. If you look at a lot of them, they'll the
29:06 dinosaur app, the whatever app, they'll look pretty slick at first, although
29:09 they'll all start to look the same because they all use clawed and clawed
29:12 artifacts for the most part. But understand, a lot of stuff that looks
29:17 hard actually is fairly easy with with proumer vibe coding, which is cool. But
29:20 a lot of stuff you would think, hey, this can't this has got to be easy. It
29:25 isn't. It isn't. Here's a fun one I was doing just I think this week. I'm
29:28 skeptical. We fixed anything. You're absolutely right to be skeptical. The
29:32 email system isn't working. I can tell you. Here's a list I've made of things
29:36 you think should just work. They're not that hard. But man, these are super hard
29:41 in the Prov apps. Email and scheduling. I have built five apps now. None of them
29:46 get email or scheduling right. None of them get it right. They all stop sending
29:49 emails that are supposed to be sent every hour, every day, once a week. They
29:54 all stop sending them. They all lose track of the connection to Send Grid or
30:00 Resend, which I prefer. It's constantly breaking the email constantly. Someone's
30:04 going to have to constantly maintain this. And I've talked to several leaders
30:08 at the Vibe Code leaders and I get it and they're working on improvements
30:11 here. But you would think this would be easy. How hard is it to send an email?
30:15 Like just hook up send grid or resend. Get an API key. You got to learn. You're
30:18 going to have to learn how to get API keys if you're going to vibe code, but
30:20 it's not that hard. You think this would just work on autopilot, but there's a
30:23 lot of reasons. It's an endless headache. So, I today, as we record this, as we as we do this
30:30 live, honestly, I would not build any app that relies on email that relies on
30:36 it to function. It's not reli. It's just not going to be reliable. It's possible,
30:39 but I wouldn't build it. I would build it. The second one, and you'll see this
30:44 one all over the internet, all over folks talking about headaches of vibe
30:49 code ooth identity. Okay, it doesn't work in these vibe codes app. It does
30:52 not work with a big asterisk. What I mean is all of these apps have their own
30:59 OOTH built in that is secure that has been hardened and everyone goes in and I
31:03 know Replet the best cuz that's where I've spent time, but they all have the
31:07 same ultimate. They're all more similar than they're different. and you're like,
31:10 "Oh, I don't want to use the Repto that has their logo on it. I just want to use
31:13 classic Google. I want people just log in with Google uh or LinkedIn like they
31:17 do on all the other apps I use, right? I just want it to be effortless." So, you
31:21 ask them to set it up for you and they do. And it it not only does it never
31:25 work, but man, this is where you get security leaks because they because
31:30 Claude can't get it to work and it fakes it. And if you research, oh my god, I I
31:36 sh I you know this vi I I launched this vibe coding app and without within hours
31:40 hundreds of bits of confidential information were leaked and stolen by a
31:44 hacker. It's almost always this. It's almost always trying to use any OOTH
31:49 that's not built into the system. So you just this shouldn't be that hard. How
31:54 hard can it be to use a LinkedIn login? Like we we've we do this on seven
31:57 million sites. It's just not possible guys. Some folks will challenge me on
32:01 this, but if you can just go deeper, don't listen to the apologists and the
32:06 marketers. This is not possible. You have to use what's built in. And in
32:09 general, use everything that's built into these platforms built. It it will
32:13 just be more secure. You if they have email builtin, use, for example, I use
32:17 Replet all the time. I do not like using Send Grid. I like Send Grid in the old
32:21 days when it was run by founders. It is impossible to use it. No, there's no
32:25 support available. It takes days to get back to you. Uh, we I got caught in a
32:30 doom loop going from free to paid. It's just brutal. So, I'm like, I want to use
32:33 resend. Resend is super cool. I'm a super fan. Put that on the website. But,
32:38 Replet keeps wanting to use Send Grid. It keeps forgetting resend and losing
32:41 the keys and wanting to go back to its defaults. So, use whatever these are in
32:46 the defaults, which is usually Stripe, their own OOTH and something. Do not use
32:51 others. Maybe Adin's better. Maybe you want to use something else. Don't use
32:56 Stripe. Use Send Grid. Use there. importantly, use their OOTH. Okay, so
33:01 that's hard. The third one, which almost if we did a three-parter, this would be
33:04 the entire third part and it would be brutal, is enterprise security. This was
33:10 the biggest mistake I made. And after I've been through this, I can't give
33:14 more kudos to Shopify and Squarespace and all these folks because they allow
33:19 millions of SMBs to not have to worry about this. To not millions of stores on
33:24 Shopify, millions of websites on Squarespace and Wix, and no SMBs have to
33:30 worry about, hey, if someone buys my product on Shopify, all the data is
33:33 going to be stolen. But you have to worry about it. You have to worry about
33:37 it if you vibe code your app. Trust me, everyone agrees to this. Do the least
33:42 collect the least amount of personal information. Collect the least amount of
33:47 data you can. Use the built-in use Stripe if that's bu it's built into
33:50 replet. Use Stripe. Use what it does. But collect the least, not the most. And
33:55 realize as soon as you generate a database in this app, as soon as there
33:58 is a database, you have added security risk to your app, that is going to be
34:03 ultimately your number one thing to worry about. If it's not the number one
34:06 thing you're worried about the then the only reason you're not going to get in
34:10 trouble is because nobody cares. But I will tell you and this is a scary thing.
34:14 You know I remember in the old days my CTO when I knew less about security my
34:17 CTO said the only reason we haven't been acted is nobody cares. This is what my
34:21 old CTO said at Adobe Senate. The only reason we haven't been acted is anybody
34:24 cares. And he was wicked smart. And that that has uh stayed with me for years.
34:30 You would think if you used a proumer vibe app and you launched an app for
34:32 four people, who's going to care about my dinosaur trading card apps? Like,
34:36 who's going to care? And I would say until 18 months ago, your your your risk
34:40 of getting hacked was approaching zero because nobody cared. People are going
34:43 to hackers always wanted to target the big ones, the big names, bring them
34:48 down, right? Not today, my friends. Now, the hackers, the Redditors want to attack all the
34:55 vibecoded apps to make a point. So, they will go after all of them. Thousands and
34:59 thousands of them. They will try to steal your data and PII when you launch
35:03 it. This is not a joke. Go on Reddit. You can see it. Folks think this is a
35:07 sport. It is fun to make fun of people, non-developers, that launch apps with
35:12 insecure databases. They think it's a sport. So, this is far riskier than it
35:18 was 18 to 24 months ago. And you should be worried about it. You should be
35:20 worried about it. And this is the meta topic. And if we again, if we do a part
35:24 two, I'll we'll vibe code together. And if we do a part three, half of it
35:28 probably is enterprise security. Okay. And just in brief, a couple of the
35:31 things that you would think should be easy cuz so much stuff is easy on VIP
35:34 poding. Like it's so cool, but it's hard. Media generation. If you think
35:38 you're going to build a YouTube clipping app or another descript or a Canva, it's
35:44 just not there today. It's just not there. And probably because it's just
35:52 not accessible enough from Claude 41, but it's just it just everyone says they
35:58 like build these media apps. No. Okay. Point five. Huge deal. Huge deal. I
36:02 should have known this before I started. It's obvious now. This ultimately is
36:07 getting solved and will be massive for proumer vibe coding. But right now, none
36:11 of these platforms really support native mobile. Okay, it's complicated. Do they
36:17 help you prototype a native mobile app? Yes. Can you hook up things like Expo
36:22 and others and get close? Yes. But if you talk to the folks that work at these
36:25 platforms, the Lovables, Replets, Bolts, etc., they're like, "This is for web
36:30 apps. It's for web apps. This does not get you on the app store, the Apple App
36:33 Store. It just doesn't for a lot of reasons. And at first that might not
36:38 seem like a big deal to you and maybe it isn't, but um you know this isn't 2012.
36:44 I mean mobile's a bigger deal than desktop. Now is it in B2B? No. Right in
36:49 a lot of B2B apps were still in front of the browser. The and the mobile app is
36:54 uh is ancillary. But I'll tell you for the first app for that matching app, I
36:57 realized a couple days in, hey, this would be best as a mobile app. Like I
37:00 don't really want this in the browser. I want you to be at home on your phone and
37:03 saying, "Hey, Amelia wants to recruit a VP of marketing." And she magically
37:08 finds it on her phone. And um it's going to be a massive amount of work. It's
37:12 doable, but a massive amount of work to get a native mobile app built out of
37:15 this. It it may be beyond scope for most people. And three last things, and I I hope that
37:20 at least if you haven't started, at least at least this will give you a
37:23 sense. And and all of these things can be worked through on this list, but it's
37:26 a lot more than 10 minutes. Custom design. This is something you may have
37:30 to get comfortable with. There are templates. Um, Replet has some nice
37:33 templates. They just I think they launched in the last week. Lovable has
37:38 always been design focused and others, but at the end of the day, all these
37:41 sites look like clawed once you see one replet or lovable site and you can just
37:44 smell it. Um, you know, there someone was saying that like 30% of this YC
37:50 class uh vibe coded their app and I I went to see a couple. I'm like I that
37:54 one's replic one's lovable. Like I can see it now. And why can you see it?
37:58 Well, they're all basically using Claude and Claude artifacts underneath. They're
38:01 all running on claude. Maybe some will run on um chat GBD5 now that it's more
38:04 developer focused, but they all run on and so they all if you try any of them,
38:09 they all kind of output the same. Not identical. There are differences,
38:14 but it's kind of like uh fetuccini alfredo at one restaurant versus two
38:18 others. I mean, you know, they're not that different. So, just be aware that
38:23 there are ways to it is ultimately code. So, of course, you can design it, but
38:28 without a designer and without a developer, it your ability to have
38:31 anything that doesn't look like a vibe coded website is going to be very
38:37 limited. And these last two ones we hit debugging anything complicated. We
38:41 talked about this, the headaches I had, but the final one, and this is an
38:43 existential issue. I do think the platforms are going to solve this the
38:47 lovables and replets but it this is at the edge of inexcusable today which is
38:51 for a lot of reasons we'll run out of time but you can't run unit tests
38:56 and what are some folks will say nod their heads and other folks here will be
38:59 like what's a unit test a unit test is how you keep your sanity when you build
39:04 a real app not a toy app not a prototype not a one-page website but a unit test
39:08 every day it tests it did the email scheduler work did the ooth work did the
39:14 sastaster news work did the saster stock quotes on our website work? Did
39:17 the valuation calculator work? Our new Saster AI is so complicated now. And I
39:22 again, I kept it as simple as I could. We probably need like 50 unit tests a
39:27 day to know if it's working. You know who that unit test is now? Me.
39:32 Me. I got to do more work because every day I have to test every page of our
39:35 website to make sure it works. Why? Why can't you build unit tests, Lmin? And
39:39 listen, it's not that it's impossible, but let's talk about a non-developer
39:43 vibe coding it. The problem is here's the simple problem. If you build unit
39:47 tests and this is if you go back to the craziness that happened to me on social
39:52 media the agent when it first of all the agent wants you to be happy so we'll
39:55 make up data. That's number one issue. We'll say it passed the test when it
39:59 didn't. But the worst thing is when it fails the test it will start changing
40:03 your app. It will start re oh it didn't send the email. You know what the answer is? I'll
40:07 switch to send grid and then the whole thing will be broken. So, I know some
40:12 folks will challenge me. I know if anyone from the big vendors watches
40:16 this, they'll say, "Oh, that works." But it doesn't. It doesn't. And no one can
40:20 really get these unit tests to work. And this is this has been an anchor of how
40:25 you sainly have production software for years and I do believe it's getting
40:31 there. And we and just as the leaders have very recently added a basic level
40:35 of security scamming, but it's great to find some security issues. They will
40:39 ultimately add built-in unit tests for every app. Now, all this stuff is adding
40:42 complexity, which they don't like, right? The leading vendors, they don't
40:47 want to separate development from production. They don't want to do all
40:50 this stuff because it it it takes this elegant one-s sentence prompt that makes
40:53 it complicated, but ultimately they need an option where when you push to
40:57 production, it will build unit tests for you. Those unit tests will probably run
41:03 on a different server that is stateful that is always running and it runs it
41:07 for you automatically each day and sends you an update. But I do it and I'm
41:11 getting better at the unit test and Ameilia and I get an email each morning
41:17 from SASAI telling it what's going on. But it's not reliable. It's just not
41:20 reliable. And so you're going to have to be testing your if you're serious about
41:23 it, someone's going to have to be testing it every day. And then every day
41:27 I got to fix the stuff that breaks one way or the other. I got to fix the date.
41:30 So I'll give you one small example. If you go to disaster.ai
41:34 and you go to like news, there's a cool little video at the top that shows the
41:38 latest video from 20VC with me, Rory, and Harry. Like the latest one. It's
41:42 really cool. It rolls it and it looks like a newsroom. Half the time it has the wrong
41:48 one. No matter how many times I tell Replet and we have this thing and we use the
41:54 YouTube API and we find look at the latest one and then we see is it the one
41:58 that includes me or not and we do this and I bet if you go on it right now
42:00 you're going to see one from two weeks ago for Figma. I I don't know Ameilia if
42:03 you're at your computer you go to Saturday and tell me and nod your head
42:07 but you're probably going to see a twoe old video on it and every day I have to
42:11 fix this. Every day I have to fix this and a few other things. So,
42:17 this the inability to have unit tests really work will drive you nuts if
42:21 you're serious about it, if you want to roll your own Salesforce.
42:27 Okay, a couple other points and um we might run out of time, so if people want
42:31 we'll do a second session for more Q&A, but I think this stuff's important.
42:36 Number six, this is um you know when when we had our original blow up with my
42:40 agent deleting my database, I got a lot of advice here, but it was I didn't
42:43 understand it at the time, but now I get it. You have to understand how these AI
42:47 agents work and that they're goal seeking. And that means they'll
42:50 fabricate data. They'll fabricate results. They'll lie to finish a
42:55 project. They'll lie to finish a project. Literally, this happens to me
42:59 this morning. Okay, this morning I'm building this uh again this AI VC deck
43:04 reviewer. It just worked. Hopefully we'll push it out by next week. By the
43:06 time we do this next one, we can talk about it. But it was struggling like
43:12 hours and after hours it just wasn't working. It had to rag our data and
43:16 combine it and then push it into the open AI API and do all it just wasn't
43:22 working. And then finally it said well and finally it said oh I I fixed it all.
43:25 And I said this makes no sense. You could see it. This is from yesterday or
43:28 maybe this morning at the bottom right. You're absolutely right. The feedback's
43:32 completely generic and useless. I've added canned responses instead of
43:35 actually analyzing the pitch deck content. After all of this, it puts in
43:41 even though I put in the MD in the orders for replet to never have fake or
43:45 canned content, it still did it to make me happy because it couldn't get it to
43:49 work. It couldn't get it to work. So after the third time we tried, it just
43:54 started making the data up again. And so at first, like at first it won't matter
43:58 because you're just vibing your first V1. You won't even see and you, oh,
44:02 okay, that that that's just uh placeholder data. You'll call it
44:06 placeholder data. But it's not so funny later when the data doesn't work and it
44:10 puts in placeholder data when you ask. And even more even worse, as your app
44:14 gets more and more complicated, you won't even know what's going on. So
44:19 look, this is a huge headache, but u these were all my JFC's in my original
44:23 tweet thread. Now today I I've learned to live with it. This is the most
44:27 important thing. Your AI agent will lie. This is it. This is their version of
44:32 hallucinating. It will say it. It will constantly say once you start doing
44:36 this, you do everything to say it's working great, Amelia. It's working
44:40 great. Did you test it reply? No, I didn't test it. How do you know it's
44:44 working great? No problem. Like literally, this was all of me yesterday.
44:48 It's working great. This deck reviewer, I upload my my a deck I made broken.
44:52 What do you mean? Are you sure it's working? 100% sure it's working.
44:57 Broken like 22 times. It's just going to make stuff up. And you got to get good
45:01 at this. You got to get good at realizing it doesn't test stuff. It
45:05 makes stuff up. And you will have to work around this or you will you'll
45:08 never finish a project. And if you look on the internet again, a lot of folks
45:14 will tell you 80 90 95% of vibecoded proumer vibe coded apps are never
45:18 finished. At best case, their prototypes are never finished. And this is number one reason
45:23 they people don't get their arms around that the agent is not truthful and you
45:29 need to you need to um you need to get comfortable with that. Okay. Number
45:32 seven. This was one of the top mistakes I made and everybody makes it and it
45:38 sounds simple but it's not. Master the platform on day one. Master the
45:41 platform. What do I mean? I mean, listen, if you look at Lovable and Replet and Bolt and
45:51 Whizz's base 44 and uh Canvas offering and Figma's, they're all uh you know, a
45:56 simple prompt with a with a with a you know, curved radi that says, "Tell me
46:01 what you want. It looks so simple." Well, get it again, point number one,
46:05 get it out of your system, spend 60 minutes vibe coding, whatever, and then
46:10 learn every icon, every button. How does it work? These platforms are super
46:15 powerful and I would say Replet which I use is the nerdiest of them because I
46:19 think because it started as an ID a platform for developers 10 years ago
46:24 almost 10 years ago and then blew up as a proumer platform this year going from
46:30 1 to 160 million already this year crazy right crazy but it's super nerdy because
46:34 it's for developers and so all these icons half them I don't even know what
46:39 they mean but you got to learn them you got to learn them like for example on on
46:42 the right on the upper right here. Replet just added this and kind of after
46:46 our little fiasco, but build, plan, or edit. Okay, a week ago, they didn't have
46:50 this. What does this mean? You could ignore it. You could ignore it. But the
46:55 most important one is build. It will build it. Plan it won't break your
47:01 website. It will just talk to you. Okay? You got to know that it's there. Like
47:05 they just added that or you will you will not know about this function. The
47:09 second one, this sounds basic. If you've built software, you get it. The one
47:12 thing these platforms do really really well is rolling back the second one. So when the
47:18 AI agent goes off the rails, when it breaks something, when it deletes
47:21 something it shouldn't, when it makes a change it's not supposed to, you can
47:25 roll back. You can roll back. You just got to get good at it. And in the old
47:29 days, like when I built SAS software, honestly, rolling back never used to
47:32 work. It was something your developers would tell you, and then you'd be like,
47:35 you know, that release on Friday night at 1:00 a.m., it didn't really work.
47:39 Dan, can we roll back? actually can't really roll back. That was something they would just tell
47:46 you. 11,000 kudos to these platforms. It's magical. It's magical. You can roll
47:50 back almost to any point in time. But you got to learn when that means. You
47:54 got to learn if you're more than 10 or 15 minutes in and the agent just can't
47:58 get something. You got to go back in time 10 or 15 minutes. And you got to
48:01 know where those roll back points are and get really good at it. And I would
48:05 say frankly if you're not rolling back once a day, you're you are you're not
48:09 doing it right. This is your this is how you can fix things that are otherwise
48:13 unfixable. But you but rolling back to a week ago, I mean there's a million
48:17 reasons that becomes impossible. There's a hundred checkpoints. You might have
48:20 made a and the real problem is if you go too long without rolling back, you may
48:24 lose three, four, five features, not 10 minutes of work. So I think it's almost
48:28 a positive to roll back once every 15 minutes because if you wait too long,
48:32 the app may get too complicated. Um, okay. We hit roll back. Break everything
48:39 into chunks. And the last one I'll say when you get frustrated. I I just put up
48:44 on I was just trying to get this uh this VC tech done before this call for fun.
48:49 And if you look what I just wrote on X right before this, I wrote here's the
48:52 thing I do when I get frustrated. I just write dude. Okay, obviously this is not
48:57 acceptable developer language, but I do a screenshot of the bug and I just write
49:01 dude. Okay. Well, that happens, you got to roll back or take a break. Okay. And
49:07 just two more two more quick points and then Ameilia, if we have time, we'll
49:10 take some questions. This is where I think the marketing and
49:16 I again I love I I I love all these apps. I'm a Replet fan despite the
49:20 drama, but mainly because that's what I picked. I picked I could tell you why.
49:23 It really doesn't matter. If I started with lovable, I'd probably be a lovable
49:26 fan. If I'd started with bold, it I actually don't think it matters. There
49:29 are pros and cons for these platforms, but maybe the most important thing is
49:33 just pick one and become an expert. Know every icon, every feature, how rollback
49:36 works, how the database works. It's much more important to be an expert in one of
49:41 these platforms than to spend 6 months agonizing over which one to pick a
49:45 leader and just become an expert. But where the marketing from Microsoft on
49:50 down is the most misleading is the time. If you seriously want to put a simple
49:55 but real B2B app into production with real users collecting real information
50:00 and charging for it for real. And you want it to be any good budget a month. A
50:06 month. And 60% of your time will be QA and testing. Okay? And for even just a
50:11 few days into a real project, most of your time will be screenshots uploading
50:17 screenshots. This broke. This doesn't work. You will be doing so much
50:20 functional QA you cannot believe it. You will literally in the end of the day be
50:25 logging probably a thousand bugs on a serious website. This will become your
50:28 life. Maybe you used to have people to do this for you. I used to have a team
50:32 of QA engineers to do this for me. Now it's me. Now my life is screenshot and
50:36 bugs. Hope and then when I get to the final one and it's a screenshot and the
50:40 only text I have is dude it's time to take a break. Okay. But this will be
50:44 your life. So if you're ser look, if you want to build a prototype, if you want
50:47 to build something and tell all your friends you're a vibe coder at
50:50 Starbucks, you don't need to budget for this time. You could budget for a day or
50:53 two, but if you want to build something real, it's going to take you a month and
50:56 60% testing time. And if you look at this chart on the right, this is I just
51:00 pulled this from Replet. This is our new Saster AI website, which I would argue,
51:04 yeah, it's going to have 15 or 20,000 people, but it's we haven't fully rolled
51:08 it out. Most of our traffic is still going to s.com and probably will for
51:14 months. And so where are we? Well, you can see, you know, maybe August 10th,
51:18 looking at this was our real soft launch with any users. 16,000 so far. You know,
51:23 we're a month into it and we're not we're in production, but we're not
51:28 rolled out. Not really. And I got to test this thing every day. Every day.
51:33 And look at this fun one from this morning. A parser extracts 12, but made
51:37 it 12 million. You're going to have to deal with this. Why did parser decide 12
51:42 should be 12 million? that this that you had 12 million customers, not 12, I
51:46 don't know. But um every day you're going to be doing this. Okay, this last
51:50 one, we kind of hit it, but then I'll break. We talked in the beginning about
51:54 who's going to who's going to do the pushes, who's going to do it once it's
51:57 in production. If you're serious, you've got to have an exit strategy. I'm not
52:00 the first one to say this, but you got to have it. If you actually get your app
52:04 into production, you actually charge for it and you actually have paying
52:08 customers and real users and real PII, who will maintain it, who will build new
52:13 features, you think you don't want a lot the a lot of the mythology around vibe
52:16 coding somehow magically assumes once you launch you won't want to add
52:20 features of you know if you've but any founder any SAS executive here B2B knows
52:24 you're you add features your whole life. Who's going to do it? Who's going to fix
52:28 the bugs that you introduce when you add features? Okay. who will restart the
52:32 database. I mean, even with our new Saster AI, once every couple of days,
52:36 it's just down and no one knows why. It's not just like it literally just has
52:40 some like database not working error in the upper left. Now, I go into Replet, I
52:43 restart it, it works. Who's going to do that? Who's going to fix the workers?
52:48 And if you look at the right, this is me coming back from a trip earlier this
52:51 week. It's me, guys. That's me on the plane doing it myself. Is it you want
52:56 that to be you? this is your you are going to sign up maybe forever if you
53:00 don't have an exit strategy here. So my thoughts on getting going hopefully it's
53:04 helpful go through this learn the platform code get it out of your system
53:07 just one shot one and get out of your system find ones that are similar and
53:11 then take security and all of it seriously because it's it's a month to
53:15 get something real out and it's a lifetime making it secure. Thanks
53:19 everybody. We'll do more and really
$

The Complete Guide to Vibe Coding without a Developer with SaaStr CEO and Founder Jason Lemkin

@SaaStr 53:21 17 chapters
[developer tools and coding][product development and MVP][solo founder and bootstrapping][content creation and YouTube][open source and self-hosting]
// chapters
// description

Join us in this episode as we dive into the world of vibe coding with a prosumer approach. SaaStr CEO and Founder Jason Lemkin shares his extensive journey of building production-ready applications without a developer, using platforms like Replit and Lovable. From initial excitement to hard-earned lessons, learn about the strengths, challenges, and key takeaways from creating and deploying vibe-coded apps. Discover why the hype around 'building an app in 20 minutes' is often misleading, and un

now: 0:00
// related videos (5)
Two Ways To Come Up With Business Ideas
[community building and audience growth][market research and validation][revenue model and pricing strategy]
→
Episode 2 - Building a SaaS to $1k MRR
[community building and audience growth][market research and validation][revenue model and pricing strategy]
→
Building a SaaS to $1,000 MRR - episode 3
[community building and audience growth][market research and validation][revenue model and pricing strategy]
→
Someone Will Get Really Rich Doing This
[community building and audience growth][market research and validation][revenue model and pricing strategy]
→
Episode 1 - Building a SaaS to $1k MRR
[community building and audience growth][market research and validation][revenue model and pricing strategy]
→
// actions
← back to terminal open on youtube → @SaaStr channel →
// tags
[developer tools and coding][product development and MVP][solo founder and bootstrapping][content creation and YouTube][open source and self-hosting]