🚀 From High School Websites to AI Aggregator: Andrei's Indie Hacker Journey 💻

[00:00:00] Today I welcome Andre on the show, so thanks for accepting my invitation.

[00:00:03] Hi, thank you for marking for posting.

[00:00:06] No problem. So if you will present yourself in two or three minutes, how you will do that?

[00:00:11] Like talking about your past and your entrepreneurship journey?

[00:00:16] Yeah, so I'm on Indie Hacker. I started building websites when I was 14 back in high school.

[00:00:25] I built WordPress websites, learned how to code, learned HTML and CSS.

[00:00:30] Then started a small web development agency after university where I learned PHP and databases and so on.

[00:00:41] That failed. That was my first startup. After that I got jobs in marketing and product

[00:00:48] while still doing side hustles. I did a bunch of things. I did everything from affiliates to

[00:00:56] WordPress blogs and a bunch of other stuff. P-commerce as well. I did

[00:01:02] tried to do a million dollar homepage for crypto at one point.

[00:01:07] Yeah, and then I got into AI when the stable diffusion boomed. Initially with

[00:01:14] an aggregator about stable diffusion. Then I noticed that a lot of products were actually not

[00:01:21] image products, they were text based. So I started there as an AI for that.

[00:01:28] Okay, great. Thanks for this intro. I will come back on certain things you did.

[00:01:35] You said you started by Indie Hacker like what you want in high school and doing some more.

[00:01:40] Yeah, before the term Indie Hacker was quite actually.

[00:01:44] Yeah, I think it was the same for me. I didn't know that it existed but I was doing it.

[00:01:50] You said something that interests me. You started a web agency and you said you failed.

[00:01:57] What was the fail in this?

[00:02:01] Many things actually. I got everything wrong. It was my first business. I initially went

[00:02:08] in with a business partner. Then turns out the business partner didn't want to put in half of

[00:02:13] the money so it was just me. Then I only had one developer and the developer left after once.

[00:02:19] So that's when I had to learn code because I had ongoing projects and I had to deliver.

[00:02:25] But yeah, also it was targeted to I actually came back from London. I started in London.

[00:02:31] I came back in Romania to start the web development agency for Romanian customers.

[00:02:35] This was the number one mistake because Romanian customers didn't have money to develop websites.

[00:02:42] I should have targeted any other Western country.

[00:02:48] Many reasons. It was an important learning step.

[00:02:54] Yeah, I believe so. Every fail is a big learning in many aspects.

[00:03:01] I'm wondering, you choose to close this agency or you really add something that forces you to close it?

[00:03:07] I close it because I ran out of money. Actually, it's funny how I got to...

[00:03:14] It was an investment of 18,000 euros and I actually set that aside over three years by

[00:03:21] having periods of around six months at the time where I lived on one pound a day.

[00:03:26] And I literally stacked the money and after three years I had about 18,000 euros and I decided to

[00:03:34] start the stages because I was passionate with building websites and stuff like that.

[00:03:40] Why did you want to have so much money? It was to pay yourself? Were you making it or...?

[00:03:46] No, to pay other people. I wasn't paying myself.

[00:03:49] Okay. Oh wow, so you didn't pay yourself this?

[00:03:52] No, no. Definitely not.

[00:03:55] That's game later in the journey of entrepreneur. We pay ourselves last.

[00:04:02] I'm seeing back my notes and I'm seeing how bad I wrote what you said. That's very funny.

[00:04:11] Okay, so there's something interesting. You say like you got your first job in kind of marketing

[00:04:18] so you didn't went for coding. Why you didn't went that way?

[00:04:21] Well, I actually started marketing. Actually, I started foreign languages before I

[00:04:28] started marketing then I went on to study marketing and that's what I like. I like both

[00:04:33] languages and marketing quite a lot. And it was... So back in the days when I started just to

[00:04:39] just remind you a little bit, it was... So the way SEO looked for example, this one thing,

[00:04:45] we had PageRank. So you had the PageRank toolbar and you could actually see your PageRank and everyone

[00:04:51] else's PageRank and you could learn how to improve it. There was like link exchanges.

[00:04:57] Yeah, so a completely different paradigm. What it was very interesting.

[00:05:02] So I kind of stumbled upon coding when I needed it, when I needed to do something.

[00:05:09] That's when I learned. For example, I learned database because I was

[00:05:14] trying to do cold outreach and I needed a place to store the contacts that I contacted.

[00:05:21] So that's how I learned SQLite. And yeah, in the beginning it was a mess. I don't want to

[00:05:28] open those old projects one. Yeah, I didn't know what they were doing. But that's how it was.

[00:05:34] Yes, definitely. But you know, even if you started your journey or you're

[00:05:39] like studying as a developer, your first project was shit as fuck. When I look back,

[00:05:44] I'm like, oh my god, why did that? Yeah, I remember I had a page where I was basically

[00:05:51] making multiple queries and I was wondering why it takes like 20 seconds to load. I couldn't

[00:05:56] have that much time to watch. Yeah. What I love the most is when you look back at your

[00:06:01] project on your code and you're like, oh, I made this and this size of the code

[00:06:08] for something. Now I know how to do it like that, you know, because you're smarter in the approach.

[00:06:14] So you don't need to have so many if, if, if, if, if. Yeah, plus now you have APIs for everything.

[00:06:21] You can use external things like people use Superbase for a for databases. I don't,

[00:06:26] I don't do that because I like building things that scale and I like to hit, you know,

[00:06:31] 100 million rows without paying the thousand dollars or something.

[00:06:36] So I use like open source databases.

[00:06:39] That's interesting. Oh, Superbase is open source. So you could host it yourself.

[00:06:43] Sure. And I do.

[00:06:45] Okay. I don't know that. Yeah.

[00:06:46] But I guess most people that use it don't use the open source. Otherwise,

[00:06:49] just go ahead and my main project, CapGo is using Superbase. And I think it cost me

[00:06:58] like now 60 euros a month. I just removed the analytics part because the analytics part was too

[00:07:05] much like it was like more than billion of row and the problem was not like the storage. It was more

[00:07:12] like the Postgre was not capable to ingest so many data by seconds. So Superbase was not a

[00:07:20] problem at neither the price. I think the price went to, I don't know, 150 200 for this.

[00:07:26] But the problem was different in this case is like they don't have a good support for analytics.

[00:07:33] The thing is if you are actually willing to learn like lower level stuff,

[00:07:40] you can get a lot done with like super cheap budget. For example, when I started

[00:07:44] for the first year, I was running on a head snare server, bare metal that I paid $45 a month for.

[00:07:54] And I was serving millions of visitors from a $45 server and the server was at like 3%.

[00:08:02] So you can get pretty far. If you go bare metal, if you go with even my scale, I use my scale.

[00:08:09] Postgre is great too. You can get super far. And PHP is, I use PHP. That's the only language

[00:08:16] back in language I know. It's super fast. So unless you're doing crazy stuff like

[00:08:22] making mistakes, you shouldn't worry about speed. For me, this is a dumb thing to worry

[00:08:31] about speed or whatever because whatever you use, it can be slow. It can be fast. It depends

[00:08:35] like how much you know what you're doing with it. And anyway, if you're building a business,

[00:08:42] your problem is getting customers and getting money, not the further speed of your website.

[00:08:47] Exactly. Like no shine was super slow for years. Yeah. And people were still using it

[00:08:52] because it was a good project. So if you do a good project, you don't need to

[00:08:56] like care about the speed until you get big enough. Exactly. Until you get big enough,

[00:09:03] 10% improvement in speed brings you actual like hard cash.

[00:09:09] Yeah, yeah. Probably never that's never happened unless you are like very, very far.

[00:09:14] Yeah. As an example, like on Cabgo, I was doing the beginning, the analytics in Postgre.

[00:09:20] And it was working well until I got like too many users and I was like,

[00:09:26] Whoa, that's quite a lot of users. Because me, I don't have like, you know,

[00:09:31] I don't have a lot of clients in Cabgo, but my client have a lot of users who call my API.

[00:09:37] Okay, guys. So that's bring millions of device every day. Like I think I am at

[00:09:44] 14 to 16 million call a day now to my API. So it's, it gets crazy.

[00:09:52] Have you tried the Mixpanel?

[00:09:56] No, I'm not using Mixpanel. I'm using LogSnag from Cheyenne where I got in the podcast

[00:10:02] to do some analytics and I used for the front end analytics, Microsoft Clarity.

[00:10:10] Oh yeah, that's great. Yeah. Make sure you have your terms of service.

[00:10:17] Yeah, but this Clarity I think I will disable it soon because like I don't want to have like

[00:10:24] terms of service and stuff like saying like I'm giving things to Microsoft. So I just put

[00:10:29] it to improve a bit the front end and I will remove it like you were supposed to stay

[00:10:32] Microsoft Clarity is an insane product and the fact that it's free and has no limits

[00:10:40] tells you something about who's the product. It's clearly brought now in the section of

[00:10:44] pricing, you go to pricing, they say it's free because we trained the IAE on your website.

[00:10:51] So it's clearly stated you're like training them. Yeah.

[00:10:55] But it's super worth it for a few things. So one of them is they have this feature called

[00:11:01] Rage Clicks. Yes. They detect when someone repeatedly clicks anywhere on the page and then you can

[00:11:08] go and view those events and see it's usually something broken or something that's

[00:11:14] more or less broken in the UX that makes the user stuck on something. Yeah, that's a

[00:11:20] super cool feature and another one is you can for debugging specific issues that specific

[00:11:27] users run into you can follow the same path that they did and see exactly where they got.

[00:11:32] Yeah replay the session. That's also a great. Yeah, yeah.

[00:11:35] Yeah, that's cool. I have a good example of the Rage Click thing happening where I

[00:11:40] truly didn't believe that would happen and people got frustrated. In CapGo you can say

[00:11:45] you can send an app update to CapGo and I will distribute the app update to all your users

[00:11:52] and this app update is linked to a channel. You have a production channel,

[00:11:56] you can have a development channel, whatever. And some users wanted to delete a bundle and they

[00:12:02] click and the UI was saying sorry this bundle is already linked to a channel so first you have to

[00:12:09] remove it from the channel before removing the bundle. I thought that was smart to say

[00:12:15] that and then I replayed the session and I saw the guy fucking Rage Click delete.

[00:12:19] It's like just read the message. No?

[00:12:25] Yeah I changed that now and you have a model saying this is linked already do you want to

[00:12:31] unlink it first? Yes. And no, no. So it's usually when something goes wrong

[00:12:36] that direction it's usually your fault as the maker of the product and some of the users fault

[00:12:41] even if like logically speaking you're right what you should you know catapult the user one.

[00:12:48] Yeah that's why I agree with you this is so useful because of course if you believe the user

[00:12:53] is wrong then you have a problem already but I never believe that and so when you see that you're

[00:12:59] like fuck I put a message but no one's care about the message you want to delete the bundle

[00:13:02] his main action is this so why am I not letting him do it? So yeah I've changed the code last

[00:13:08] week I push it in production and now I don't have Rage Click anymore. Yeah nice, nice.

[00:13:14] Yeah you really notice these things and you notice that the users don't read stuff they don't

[00:13:22] read labels they don't text they just you know they read the UI so it's that's what they read

[00:13:28] they are used to certain patterns in the UI they look for the patterns that they know

[00:13:32] if they don't find them yeah that's when they get blocked. Yes I had a funny case

[00:13:41] like that yesterday I think someone contacted me on the support where the support is basically one

[00:13:46] click away from the same place where the user wanted to do the things and he sent me a message

[00:13:50] can you delete my account I don't find where he is. Next button it was like profile and then

[00:13:56] you see it in front of you you are that close and I'm like okay yeah sometimes this one

[00:14:03] I don't know how to fix it better but yeah I don't want to say it's mostly German users but

[00:14:15] possibly possibly no I want to check for

[00:14:21] okay let's go back to your thing because that was we deviated a bit so you you said like

[00:14:28] when you got your first job you started the side assault again like during affiliation

[00:14:34] e-commerce did you had any success on that? Modern success I didn't sell

[00:14:41] okay then I got to sell my business or anything I just rolled it until they died

[00:14:46] rolled which project I believe but okay yeah they were side projects for me

[00:14:53] so I was they weren't my main source or become what's I did games as well I did anything that

[00:15:00] gave to mind you know yeah I did there's I think it's still up it should still be up

[00:15:08] it's dinosaur game let's see dinosaur game.net it's a multiplayer version of the chrome

[00:15:15] offline dinosaur game where you have to jump out no way yeah you made a race crazy

[00:15:22] it's not a race it's a or you could view it as a race it's basically you are competing by country

[00:15:28] I was inspired by a website called pop cat dot click where you tap the screen and it just

[00:15:36] increases the score and the scores are grouped by country and you can see there's a leaderboard

[00:15:41] of countries so I did basically yeah I did the same thing but for a game

[00:15:47] okay yeah I can make some money it makes like I don't know like a hundred dollars a month

[00:15:53] okay great that was my question and you say it was a moderate success on the affiliate

[00:15:59] and e-commerce so what is moderate for you like what was it making or did it make overall

[00:16:05] I think the most I made like as a side hustle what I have probably months when I made

[00:16:12] like 10 to 12k but like for a few months that's it what usually like around the average was around

[00:16:20] like between one and two thousand like a month yeah it's pretty cool already for a side gig

[00:16:27] yeah it was like vacation money yeah yeah definitely it's like okay let's get in vacation for my

[00:16:36] yeah let's build a game actually for the dinosaur game is actually the only one where I

[00:16:42] got an acquisition offer from from a big company I won't name the company but it's a really big

[00:16:48] gaming company that wanted to acquire it but we didn't agree on price and how did you make the

[00:16:55] money with the game by the way through ads okay okay okay super um you you talk about the the

[00:17:08] crypto lending the one million dollar crypto that was the same idea as uh I don't uh mark

[00:17:15] I forgot his name properly to say but it's just a landing when people pay and it's a ranking

[00:17:22] or so it was the website is not online anymore it was based on the idea of the million dollar

[00:17:30] home page which I'm sure many people are familiar with where you sell a pixel for a

[00:17:34] dollar so you end up selling making a million dollars yeah I basically did the same thing

[00:17:39] for crypto so you could basically promote your coin or whatever it was and it was a limited

[00:17:44] number of pixels that you could buy and you could buy it with crypto what it did uh I don't I don't

[00:17:51] know I think I got bored with it at some point I didn't even launch it it made like zero oh okay

[00:18:01] like why why you never reached to launch because you are adding more and more features or

[00:18:08] I'm trying to remember exactly I think it was um I I uh I started working on it during the crypto boom

[00:18:17] of 2017 or something like that uh and then Bitcoin crash I think so that might have been it oh yeah

[00:18:26] yeah definitely yeah I also like got bored with you know it's not my I thought it would be fun

[00:18:32] you know to advertise the crypto space and so on but it's not the ecosystem is not

[00:18:38] not very fun and also the like the prices for advertising I looked up you know a bunch of

[00:18:45] blogs where I could have promoted the song they were charging like 20 000 dollars a month for a

[00:18:51] banner it was wow crazy prices yeah yeah yeah so no way you can make it work with like any kind

[00:18:59] of you would need like 500 conversion rate to make it work it doesn't make any sense yes definitely

[00:19:06] okay um so let's go now to the project you're working on now can you talk a bit about it and name

[00:19:12] it because you earlier you didn't even give the name I think okay so a little bit of backstory

[00:19:17] because it's the the journey is actually interesting so I'm I'm a reddit and hacker news

[00:19:26] browser I got ready to hack and use every day I've been doing it for a long time

[00:19:31] and at some point a stable diffusion was launched and okay that was very cool and then

[00:19:39] for a while after a stable fusion was launched every day hacker news was full of stable diffusion

[00:19:46] based projects like everyone was building their mini project based on stable fusion and launching

[00:19:52] it on hacker news I have I had never seen anything like that in the history of hacker news like a

[00:19:57] topic dominating for for so long I think it dominated for weeks to the point where people

[00:20:03] I think they asked the admin to stop oh yeah yeah because they weren't stopping so I looked up

[00:20:12] I initially wanted to make a list just for myself for research and I discovered that there was

[00:20:18] already there was one list it was in a reddit post I don't have the reddit post anymore I think it was

[00:20:25] deleted but anyway there was no website so I decided to this time I got inspired by layoffs.fyi which

[00:20:34] is another great website about tech layoffs it's so simple it's just or last time I looked into

[00:20:40] it it was just air table embed so air table is a spreadsheet cloud spreadsheet and you can

[00:20:48] embed it in a web page so I did the same thing I collected the I think around 700 links to stable

[00:20:55] diffusion apps put them in an air table and embedded air table on a blank page there was

[00:21:02] just one page and that was it and underneath I think above I had something like a sign up

[00:21:08] for email updates and below that I had the submit your app so people started I launched it on

[00:21:15] I don't even remember where I think reddit and a bunch of few other places

[00:21:20] and it got decent traction and I got a lot of submissions so people started submitting their

[00:21:26] apps however they started submitting text apps instead of image apps and I couldn't publish

[00:21:34] them in the spreadsheet in the main list so I made another spreadsheet with the text apps

[00:21:40] and back then it was GPT2 so they were based on GPT2 yeah yeah that's how early it was

[00:21:49] so I made a list of GPT2 apps and a bunch of other like different AI tools that weren't

[00:21:54] based on GPT at all and when I got to 700 again I bought a bunch of domains like I had

[00:22:01] the like a period of brainstorming where I thought let's see what the domain I think

[00:22:06] the domain is pretty important for stuff like this if you notice that all the domains I told

[00:22:12] like layoffs.fyi and so on the million dollar complete the domain is actually the brand

[00:22:17] so I wanted a domain that's that's a brand and I initially my first I had I had a bunch of ideas

[00:22:24] that I now realize would never work one of them was machine tasks.com

[00:22:32] because I realized yeah like machine learning

[00:22:37] I see where it's coming from but it would have never succeeded

[00:22:44] I'm not sure actually maybe probably more difficult but yeah for sure more definitely

[00:22:50] more difficult in terms of virality but I had this like notion of tasks that wasn't popular

[00:22:58] at all back then I remember I googled like AI tasks no one was almost no one was talking about

[00:23:03] AI tasks not not the way everyone is talking about them right now so it was the first website

[00:23:08] that was really put an emphasis on tasks because from a consumer standpoint you usually have

[00:23:17] a task that you want to accomplish and you don't really care that much which AI tool is

[00:23:22] going to help you achieve that this is my belief of course there are etc you can build

[00:23:28] you know you can build a huge brand you know I mean if I have to do some design I will probably use

[00:23:35] camber but for most tasks I don't care I just just give me the the best tool for this task

[00:23:45] so I settled on there's an AI for that and I launched it early December 2022

[00:23:55] and it was just one page it was an index.php file that's it it had no pages you yeah it had no pages

[00:24:04] you couldn't like click on an AI and go to that AI page no it was a list of links with name

[00:24:10] AI name and task that's it Peter just did that two or three days ago with an airline list or something

[00:24:17] like that that's great that's how you start in fact what I actually um I discovered I remember when I

[00:24:24] discovered him uh because um it was this guy building websites and he was building them

[00:24:31] inside the index.php file which is the same thing I was doing I was like wow I'm not the only

[00:24:37] one in the world and it seemed pretty crazy to me even back then but it really works for testing

[00:24:43] quick ideas obviously now the website is bigger than a one page anyway so I launched it and in

[00:24:51] the first week or so it had around 100,000 visits so it got immediately popular yeah right off the

[00:24:58] back and it wasn't even the first aggregator there were a few other ones because I waited

[00:25:06] a long time to gather the list to figure out what domain I'm going to use to figure out a bunch of

[00:25:12] like strategic things but I was I think the third or something like that all the sites like Taft

[00:25:21] I'm calling it Taft that's like the short name

[00:25:25] all the sites that launched in 2022 are still around and are decently popular

[00:25:30] the sites that launched after 2023 usually not uh not that much around because right now no one looks

[00:25:39] for new aggregators I mean that that ship has probably is going to be a new thing in the future

[00:25:48] yeah or a new topic but not not in actual one like when one is working it's like difficult

[00:25:56] yeah what it's so I find it really fascinating that you can build big businesses around simple

[00:26:04] websites with not a lot of tech you would seem like for a for a big successful business you need to

[00:26:11] you know build some groundbreaking piece of technology that's actually the hard path

[00:26:17] because then you need to educate people on why that's better it has to be way better it has to

[00:26:27] it's usually easier to launch a single feature that doesn't exist yet so you're comparing it to

[00:26:32] naffi so it's infinitely better uh and I was fascinated by a coin market cap because it's

[00:26:42] I remember like the days of coin market cap it was a website between you could tell that it

[00:26:48] was built with bootstraps yes that's pretty much that everybody uses that super easy to use

[00:26:52] it was one page with bootstrap it was a table of coins and their market caps that's it that was the

[00:27:00] first major feature you have a list of coins and their market cap coin market cap and then obviously

[00:27:06] it evolved but I wanted to build the coin market cap of whatever the next big paradigm shift was

[00:27:12] going to be so I have been thinking about this for years what's the next thing and I thought it

[00:27:19] was I thought it would be AR because I looked at apple patterns and a bunch of other information

[00:27:27] and it looked like apple was about to release something along those lines right now basically

[00:27:35] around this time which they did but still I have no idea how I would design such a thing I mean

[00:27:42] would it be a standard website would it be something that you access within AR but then

[00:27:46] how do you browse the other websites that are not AR yet yeah it was a huge problem that I

[00:27:53] went on I realized that you know in AR there are probably bigger opportunities like imagine having

[00:28:02] even entertainer wise like imagine having a real real I mean a pokemon next to you like

[00:28:09] walking around with you and so on I mean for kids it's amazing and it unlocks a lot more

[00:28:15] interesting opportunities than just educators but then you know I got lucky with the stable

[00:28:21] diffusion thing and I never saw it coming and it was kind of fooling me so that that's how I went

[00:28:28] this past and but I don't regret it I think some of the best ideas they they polluting

[00:28:33] I did not expect this would become a business not at all so I expected this would be a side

[00:28:39] project and my goal was my like ultimate goal was for this to make $10,000 a month passively

[00:28:47] so for me like if you ask uh when the moment I launched it you know I would have signed a deal

[00:28:54] that you know I would have you know given it away basically sold it in exchange for 10k a month

[00:29:00] passing uh yeah but where are you with numbers now I'm fired and that I haven't made them

[00:29:08] public yet but significantly higher nice great job and that's your main main work now you're working

[00:29:16] full time right now yes I'm working full time I have a team so I'm not a soloprenor anymore I started

[00:29:21] it soloprenor in fact I think this is a problem mistake that many people did including myself

[00:29:26] like scaling too late not not scaling when when you have huge demand

[00:29:34] uh because you're afraid because maybe you've had paid startups like I did so you know what it means

[00:29:41] to you know fire people or even yeah it sucks the whole thing failure sucks really bad

[00:29:49] um but you have to you know get over it and you have to know when it's time to scale again

[00:29:55] I think that's that's a common mistake how many people do you have in the team right now

[00:30:02] uh I think we're seven or eight full time wow that's crazy

[00:30:10] and uh yeah so well I'm not like I'm not in a rush to hire or anything I just I hire what I forced

[00:30:18] like um yes that's the best for me yeah the first hire was uh for the newsletter so

[00:30:26] just to rewind a little bit so I launched the website and everyone immediately started asking

[00:30:31] for a newsletter it was the letters were starting to pull uh and I was like I hate writing newsletters

[00:30:38] and it's just me and I'm anyway doing a bunch of things I hate like I have to do accounting

[00:30:43] well I was doing accounting by myself and all the financials and I hate that like how many things

[00:30:48] am I going to do that I hate uh and I so I postponed it plus I had an excuse which was

[00:30:53] the fact that you know I thought if this were to become a reference site for the AI industry

[00:30:59] I want people to go and visit the website every time I don't want them to uh sign up with their

[00:31:04] email which gives themselves permission to not look at the website anymore you know become

[00:31:09] passive about it I wanted them to be active about it then come back and get direct traffic and so

[00:31:14] on so I was I postponed this for like six to eight months which is crazy because I missed on a lot

[00:31:21] of emails uh until uh then I realized look at least some people want to get subscribed so what I could

[00:31:30] do is send them to another AI newsletter there was a bunch of other AI newsletters like Benz

[00:31:36] yeah so I'm giving so I please tell me you didn't give them for free

[00:31:42] um I actually wanted to but I got no reply from anyone so I contacted every major

[00:31:50] AI newsletter and back then they had like probably all of them had around like 50,000 subs so it was

[00:31:57] you know they were growing they were making money they were making way more money than me I was making

[00:32:02] very little money back then um and uh I got no replies like absolutely I followed up to make sure

[00:32:10] they got the company says someone else yeah thank god yeah uh and then someone else told me like

[00:32:18] look dude I think uh you know you should just start the newsletter and pay someone to write it or

[00:32:24] whatever and I was like okay let me try something uh so I made a script like a series of back-to-back

[00:32:32] prompts basically a pipeline uh to generate the newsletter with chat GPT and it was the first

[00:32:40] newsletter 100% generated with chat GPT so I didn't write the word of that for the first

[00:32:46] many months yeah and uh but also that wasn't great I mean the newsletter was decent the quality was

[00:32:55] okay it wasn't but I still didn't like to do that because we still have to check if you have to

[00:33:00] click the send button which is the most stressful moment when sending as I'm sure you know

[00:33:06] I was always when you start send and then you're like oh fuck I forgot this link

[00:33:11] oh yeah no I'm not sure which which provider was it that they had like uh

[00:33:18] ah yeah um I think it's some main chimp main chimp they have the the things yeah yeah

[00:33:23] they showed you your stress level when you click on the yeah yeah crazy the shaking hand

[00:33:31] yeah so I eventually got someone to take care of the newsletter

[00:33:35] and we're still improving it I think it's great but within three months or when I

[00:33:39] stopped collecting emails and I initially just launched a wait list for the newsletter

[00:33:44] I still I was hoping that no one would subscribe so I said okay I'm gonna launch a web wait list and

[00:33:50] you know if I don't get enough interest I won't launch it and I got within three months I got uh

[00:33:56] it was the number one AI newsletter like I got it was bigger than Ben's bytes within like three

[00:34:01] to four months something like that um yeah so I have to believe but you know uh with hindsight now

[00:34:10] I know that for example one of the biggest drivers of people coming back to the website is the newsletter

[00:34:19] we always feel wrong when we start things you know like because we try to rest our

[00:34:23] shelf like to not do things we don't like and at the end you do it and you're like

[00:34:27] fuck that was such a big mistake to don't do it earlier yeah yeah really good because right now it's a

[00:34:35] almost it's at 800,000 subscribers which is the same amount yeah and it's growing by a lot

[00:34:42] each month so it's a it's a business in itself right now it's a media business in itself

[00:34:49] so yeah these letters are great and I think beehive uh is uh partially definitely partially

[00:34:56] responsible for uh the researchers of newsletters what they did by lowering the price drastically

[00:35:04] lowering the price of sending them it's like try doing uh you know sending to a list of a million

[00:35:09] people on mailchip like how much are you gonna pay I don't know insane amounts a February the price

[00:35:16] there is like crazy yeah I don't even know where their enterprise plan starts but maybe like

[00:35:22] yeah anyway so I think they find they did lower it a bit after beehive like kill the price

[00:35:29] for sure yeah things but it's still more expensive but I mean beehive was uh insane insane I was

[00:35:36] initially I was skeptical I was like what's this like weird domain whatever but it doesn't matter

[00:35:40] people subscribe to it that's fine you can obviously use a custom domain so you don't have

[00:35:45] to use their domain and the features are great they're moving super fast you're still using

[00:35:50] just an amazing yeah yeah and I have no plans of abandoning them I'm like loyal to whoever delivers

[00:35:57] if they deliver like super high value even if they become slightly more expensive than the competition

[00:36:02] I hope they will not listen to podcasts it's fine they won't they won't because they know better

[00:36:11] they I think they're focused on on other things right so one of the things that

[00:36:15] it looks like they're focused on which a lot of people are focused on is expanding horizontally

[00:36:23] because the development development costs have decreased so much because of AI because you can

[00:36:28] code faster you don't have to I mean you can code in a if you are a programmer right now you

[00:36:33] can code in a language that you don't know really uh and worse yeah yeah I don't know shit about

[00:36:43] a graph tool and I will never learn it because I don't like people and I I did graph your

[00:36:49] rich eGPT just fine for some api and so on you can you don't need to know a regular expression

[00:36:54] some field that you'll never have to learn again yeah a lot of things got in there yeah you

[00:37:04] know funnily this week I got kind of a ha ha moment with the IA stuff because like

[00:37:10] long time ago I did a Chrome extension and I knew there was painful you know set up in the things

[00:37:17] be able to have things working in in Chrome extension it took me like a while to understand

[00:37:22] how to do it and lately I had an idea of something that can help myself and probably my users as

[00:37:28] well for it's super simple uh Chrome extension so it's like okay maybe I will try to do it with

[00:37:34] Claude uh what's his name the 3.5 sonnet yeah yes sonnet and I was like okay let's try that so I

[00:37:40] made a prompt like give me a plan to do a Chrome extension doing that and he gave me the plan

[00:37:45] I said okay execute the plan now and step by step I just copy paste copy paste copy paste

[00:37:49] I told him like I prefer to have a type script I want to have a tile win and stuff like that

[00:37:54] and he just did the extension he wants some bugs I sent like I just screenshot the bugs

[00:38:00] send it back he fixed the bugs and the extension is pending grubly's right now and it was full

[00:38:05] charge gpt I didn't put one line of code that was like crazy. Claude sonnet 3.5 is amazing I'm waiting for

[00:38:12] the height of the small model but it's amazing it one one shots a lot of problems it's way better

[00:38:20] and I'm sorry say this but it's way better than the gpt 4 and way better it's not close for

[00:38:26] it blows it out of the point and by the way Chrome extensions are amazing amazing

[00:38:35] like distribution platform because people rarely uninstall extensions once they install extensions

[00:38:44] the mutation is great and you don't have a lot of competition I don't want to like if you want

[00:38:51] to stop just then no continue but I know it's a good business yeah yeah it's a great business

[00:39:00] and also what I find very nice for example in Chrome extension is you basically can improve any

[00:39:08] business like if you find Notion as a problem and people complain about Notion you just create

[00:39:14] an extension for that so you don't need to create a new business you're just fixing it

[00:39:18] kind of yes you can make it paid you can piggyback on distribution channels that are not

[00:39:25] officially recognized as distribution channels meaning any apps any website that has over

[00:39:32] one million people or whatever that don't have integrations don't have any other way of piggybacking

[00:39:39] on their distribution basically however you are subject to the terms of service of that

[00:39:43] website and this is what happens happened to a bunch of tools for LinkedIn for example so

[00:39:50] from extensions that you know interacting with new dean yeah they got cease and desist letters from

[00:39:57] LinkedIn because it's operating on the website so it's still subject to the website's terms of

[00:40:03] service but Notion is well known for like playing super nice with developers even those building

[00:40:10] on unofficial stuff APIs or so I think with Notion it's safe but not with everyone so you've got to

[00:40:17] be careful there but any distribution channel has a risk of getting kicked out right like social

[00:40:22] media has certain changes have with you know every for me the point is if you can do it easily

[00:40:29] quick and dirty and try and see if you have a market or not then Chrome extension is very fast

[00:40:35] for that because you have a lot of less things to build and for example I did that for LinkedIn I did

[00:40:41] an analytic solution for LinkedIn in Chrome extension and I didn't get a letter because

[00:40:47] I read the term and service and they were saying like if you automatize things for the users

[00:40:52] then it comes less to the website we are against you but if you bring the user more on

[00:40:57] the website we are for you so what I so what I did is instead of building the analytics and

[00:41:02] put him in my website I put it in the LinkedIn website so people call more to watch the analytics

[00:41:08] and they're like they didn't block that they were like yeah thanks improve the

[00:41:13] improve the analytics so you basically you basically just build you build a feature

[00:41:17] no an extra feature yes yes and made it paid I didn't make a lot of money for that because

[00:41:27] basically building analytics on top of LinkedIn this is such a mess like they code I never saw that

[00:41:34] bad something that bad at that scale like they yeah I don't know did some things happen to be in CSS

[00:41:42] analytics side I'm like what the fuck guys what's what's happening to scrap the website that was

[00:41:47] the worst experience ever so I shut down the project because like every week everything was breaking

[00:41:53] up with the analytics impossible to get from api or the data you had to open a web browser to get it

[00:42:01] yeah this happens when dealing with building on unofficial API

[00:42:07] but official API is great too so I'm not I'm actually on I'm not on either side here

[00:42:15] think it's okay to build an official API and sometimes you know just by building on top of

[00:42:20] something you can happen to have a great business as well was lasting like I have a friend he made

[00:42:25] he made a tool for a google task you know the google when you have an android phone you have a

[00:42:33] app google task but you can put like some task in it and they don't have a desktop app so my

[00:42:40] friend made at the beginning a Chrome extension I think to have your task in your Chrome extension

[00:42:45] and then he built a website which is the google task web version which is not made by google and

[00:42:52] he made it look like google yeah the only only point where they have it it's in google calendar

[00:42:59] as a kind of module it's not well but so that's separate from google keep right yes yes that's

[00:43:05] separate and yeah and so this website now is task board it looks like google ui everything

[00:43:13] looked like google people paying they believe it's google also and you make good money for that and it

[00:43:19] still exists since long google is not killing it they were scared to be get killed after every

[00:43:25] month but still existing and does it have paid option is it monetized yeah it's monetized

[00:43:33] it's my good money nice nice nice I was I was surprised so you have ways to build in top

[00:43:41] of that so now they are building a lot of chrome extension for google docs I think now they're making

[00:43:47] like a google docs extension for to put charge epiti in it but more free than the one extension

[00:43:54] is making by google so they're doing things like that and they keep like just building in top of

[00:43:59] google like fixing every part of the product of google and they make good money for that

[00:44:04] yeah it makes sense I mean you're delivering value that's what matters

[00:44:09] yeah and and that's I think I mean you have already a lot of users in google so then you

[00:44:16] just take a part of that and you market it well it's like it's easier than what you were saying

[00:44:21] earlier like create a new things and people love to learn about it how to use it blah blah blah

[00:44:26] this you have to pick it back on something so if you're like starting from zero and

[00:44:32] you're not using any kind of distribution channel or anything idea something you know like there are

[00:44:39] these apps right now that are AI apps around looks maxing that's a like a trend for Gen Z I don't

[00:44:47] know about it okay no to all that okay yeah same what I heard about it you know so it looks

[00:44:55] maxing is just maximizing your looks and it uses AI to give you advice for skincare stuff like

[00:45:01] okay yeah so that's piggybacking so the app is called I think it looks maxing

[00:45:07] but it's piggybacking on a popular concept a popular trend you know you imagine if they

[00:45:15] had start they had done the same app looks maxing but looks maxing wasn't a thing there is the

[00:45:21] same app it delivers the same value same identical value but it's not a thing yet

[00:45:28] would it make I think it was I don't know how much exactly it was make about millions talking a lot of

[00:45:34] money yeah it has to be a thing you know I've done that many times like being too early in America

[00:45:40] like you build the things and you know like oh that would be the next thing and and then you

[00:45:47] do it but like it doesn't resonate well because you're too early and then you stop it and when

[00:45:51] you stop it they start the trend starting fuck yeah I mean there is one caveat here which is that it looks

[00:46:01] like people are willing to try apps again now so people haven't been willing to try a lot of apps

[00:46:10] since apps were a thing were new basically like in the 2010s when the apps are new

[00:46:17] there was you know there were websites that aggregated apps similar to what I'm doing

[00:46:22] and then people lost interest you know okay it's an act people stop looking for new apps

[00:46:28] I remember back then I was like browsing the app store looking for apps you know

[00:46:32] there were I was trying a lot of apps oh yeah you were using that before yeah yeah yeah yeah

[00:46:37] and then a period of like 10 years when people didn't care anymore and the only way to acquire

[00:46:43] and there was you know using the distribution channel like apps or organic you know 30s and so on

[00:46:50] or viral but now again it looks like very much like people are trying a lot of apps again with AI

[00:46:59] so they're trying it looks like rotation is average rotation is around three months for AI apps

[00:47:06] so after three months you should expect to lose your customer on average I mean

[00:47:12] it gets better so it's an opportunity and there are a lot of early adopters right now

[00:47:20] everyone is using these tools and is trying it out it's super international the phenomenon is a global

[00:47:27] one you see I mean I'm looking at the analytics and I see all countries it's incredible I've never

[00:47:33] seen I mean I've looked at you know hundreds of analytics I've never seen anything close to this

[00:47:38] like it's all the world is using this stuff yeah no no it's it's it's crazy and sometimes I even got

[00:47:45] like very funny you know CapGo my tool is open source and I am often posting tasks you can complete

[00:47:52] for I don't know 50 bucks 100 bucks or something I don't want to do I know it's not long so I put

[00:47:58] Bounty in it and some people take it and this week I got a guy he sent like two two tasks

[00:48:05] because he was like possibly paid and and I look at the at the code he sent and he looked off you know

[00:48:12] like I was like that looked weird and and then I realized like something went not like done properly

[00:48:19] I could see in the code like the code was not working and I was like what the fuck and I asked

[00:48:23] the guy did you test it and and and after a few message I understood like he don't know to code

[00:48:29] he don't know how to run code he just tried to use jgpt to get some bucks to get a new

[00:48:34] computer he told me and I was like okay wow that's crazy and the guy from India or Pakistan or whatever

[00:48:40] then I was so impressed like the guy don't know to code is like okay I can find and I can get money

[00:48:45] by using charge bt so I will do it's like okay you can try at least yeah I mean if you are

[00:48:51] willing to with cloud I think it's even easier right now so if you are willing to just

[00:48:56] copy paste the error it's back into the bot basically back into charge you can probably

[00:49:03] if you have enough persistence within a day I think you would ship a small app

[00:49:09] yeah I think what anyone can do what is missing right now is the for this guy he couldn't

[00:49:15] understand how to run the code and to have a development so yeah so it was stuck but otherwise

[00:49:22] I think he would have completed the task because he was highly motivated and that's crazy

[00:49:29] these for example cloud can also run code and it runs pretty sure it runs react as well

[00:49:36] they have a sandbox where they run a bunch of frameworks and a bunch of stuff so it's you know

[00:49:42] how much additional effort would it be for them to allow you to publish this to external domain

[00:49:49] probably not that much I saw a website not long ago who was doing that like he used

[00:49:57] cloud and he published a website of something you ask for it and he published directly a website

[00:50:01] from what you asked for wait I will try to find it back the name was a webseam.ie I don't know if you

[00:50:10] know it sounds familiar I send it in the chat and for people if they want to see as well I

[00:50:19] would put it and so this is basically a website and when you go to it it look like it's a

[00:50:26] web browser and on top of it you have like a search bar and if you put like whatever you want like I

[00:50:33] put like I tried because I have an idea lately of a music game so I put the name of the music

[00:50:41] the pitch of the music game he did it like instant almost he released it and I could kind of

[00:50:48] play with the button and stuff I was like so impressed and then he has a URL it's really

[00:50:54] very impressive yeah and you're right I think more and more things will happen like that people

[00:51:00] who can just like broad day idea and it will be like released instantly so your parent my

[00:51:06] parent will be able to do a website. I really believe that I think of course there is a chance

[00:51:12] I'm wrong and there is also an element of the future shape so we actually also get to decide

[00:51:20] this happened a lot but I think there was a chance that there will be probably like millions or hundreds

[00:51:26] of millions of AI tools built by anyone and if you look at what happened with blogs before

[00:51:34] WordPress and blogger so you could still make create a blog before those platforms existed

[00:51:41] but you have to know how to code you have to you know have some hosting know how to set

[00:51:46] that up not many people did but then you know blog spot and WordPress came along and we went to

[00:51:54] 100 million blogs overnight and this is because they made it super easy for anyone to create a

[00:52:01] website and I think something like this might happen for AI tools because some of these tools

[00:52:06] let's be honest are rappers they're simple rappers one prompt they rap the prompt

[00:52:15] and they just send a request to the T or whatever AI model they use and this is like really simple

[00:52:22] stuff and you can build useful things with those rappers I'm not saying they're useless they're

[00:52:27] very useful because they yeah they're rappers on intelligence I have a good friend of mine

[00:52:35] in France he's he doesn't know how to code I think maybe now he's learning and he do no codes since

[00:52:42] long he's freelance in no code and stuff and he was helping me with my previous podcast in France

[00:52:48] he was doing the resume of the episode before whisper and charge api exists he was doing that

[00:52:54] like I was he was listening to my podcast so he was liking that and he said yeah I can

[00:52:58] write the resume and you pay me a bit so we worked like that until I finished the

[00:53:04] I stopped the podcast in France and then he was like oh maybe now with charge api and whisper I could

[00:53:09] do that and I was like warm and this seems so complex like even for me as a developer I'm like

[00:53:15] it seems very hard to do but he did it he has a website now where you can register subscribe

[00:53:21] and translate you get the transcript of your podcast get a resume get like a title idea

[00:53:27] and stuff like that and then just in full no code and charge api I was like very impressive

[00:53:34] yeah well yeah yeah yeah whisper is amazing I mean and I think whisper is gonna be running

[00:53:40] everywhere something like whisper it's exactly the type of tool that makes total sense to run

[00:53:46] locally it doesn't make a lot of sense to not run locally because you don't it also makes

[00:53:52] from a privacy perspective you know if you have like voice notes stuff like that you don't want to

[00:53:57] send them to a server also I think for the speed it will be better to have locally like

[00:54:05] fast response of the of the thing oh yeah this is the future with I will be crazy

[00:54:12] we're not really a thing for that even I was like very impressed not so long we go by

[00:54:16] sooner like the fucking me build a music in two seconds now like and very good music like

[00:54:23] yeah yeah that was unexpected for me I remember there was this one of my favorite podcasts

[00:54:32] that I recommend every one of you if you haven't already is uh Naval on Joe Roga it's amazing

[00:54:39] have you seen it no no I didn't listen yet but it's all like

[00:54:46] four or five years old but it's super good it's full of wisdom he has this famous thread of how to

[00:54:52] get rich without getting lucky talk about it yeah there's a lot of ideas and one of the however

[00:54:57] one of the things where he was wrong perhaps the only thing where he was at least partially wrong

[00:55:03] I would argue was regarding AI so he was like he was saying like you don't need to worry

[00:55:09] about it it won't take any of our jobs not within our lifetimes it's we're nowhere near

[00:55:16] it won't be able to deploy tree it won't be able to do art we're all gonna be sitting around the

[00:55:21] writing poems and drawing paintings funnily all the artfix are doing make better by you know

[00:55:31] I mean I'm willing to admit that GPT-4 is significantly better than me at poetry

[00:55:41] how for sure thank you and many other things as well but poetry and even painting I mean Dolly

[00:55:48] I'm terrible at it so it's already better than me at these things and I by far it's better than

[00:55:58] better so yeah and I reply to a tweet of his and saying this and he blocked me

[00:56:11] it got like it got like a thousand likes or something like remember yeah because he's

[00:56:19] still like skeptical but anyway other than that other than the specific thing about it also

[00:56:29] he was just wrong about AI I mean we're all wrong but that's funny he blocked you

[00:56:39] I mean yeah I think I actually remember what happened is I replied I couldn't you cannot summarize

[00:56:50] all of this in a tweet what I just told you so instead of doing that I wrote you said the

[00:56:55] AI is never gonna replace artists or something like that and look at what happened

[00:57:01] and he replied you don't even know what an artist is like okay I like as if you know it's sort of a

[00:57:11] straw man of my argument but Twitter makes it super easy to make straw man because you're limited

[00:57:16] to a few characters so obviously anything you type is a straw man or of your own idea let alone

[00:57:21] what other people think about it but yeah that's pretty funny I think we did a bunch of things on

[00:57:29] your project now I would like to know like what's the future of your current project and do you

[00:57:34] have other things you're building I have no other things the way the way it works is I don't

[00:57:42] I try not to have any super long-term plans regarding theft and EMAI in general because I noticed

[00:57:50] that things change incredibly fast so most if I embark on a journey or that makes assumptions

[00:57:58] over assumptions that I you know that culminates in a great amazing feature that changed the world

[00:58:04] I think the chances of that not working out are very high so what I do instead is I try to

[00:58:12] figure out first of all what are the things that are still gonna be true in like the few years

[00:58:19] like I'll give an example I think does will open AI still have an API in five years I think so

[00:58:26] I think they will so some people say yeah they're gonna eat eat everything yes eat all software

[00:58:31] do you think they're gonna have an API in five years if the answer is yes then you can build on

[00:58:37] top of these APIs of these models if the answer is no when what do we really think it's gonna

[00:58:43] eat software and you know the FPC is gonna just say yeah that's fine you don't need to offer an API

[00:58:50] just do all software on your end I don't think that's gonna happen and plus their current business

[00:58:56] model needs are based on setting their API so that's one thing that that's still gonna be true

[00:59:03] I think and I try to get hints from how the way people use that so one of the advantages

[00:59:15] of having users is that you get data if you listen carefully to the data you don't have to

[00:59:22] do exactly what the users want but you have to do going the direction where it looks like the

[00:59:29] users are the behaviors yeah so this is what I'm trying to do obviously I have some you know

[00:59:37] features that I'm developing but I'm gonna wait and see there are many opportunities you know I could

[00:59:45] grow into you could grow into quite a large website over time do you use iA yourself in the

[00:59:54] because you were saying before the newsletter was made with iA and now what is using iA in your tool

[01:00:00] so I use AI for search that's uses embeddings obviously I use AI to generate descriptions

[01:00:08] to generate FAQs what else I use I have an internal agent that I made over a weekend

[01:00:17] just to see what kind of agent I can use I think it's pretty good so I tied it

[01:00:24] I give it a few set of actions that we can perform like grep search the entire code base

[01:00:31] connect list all the tables in the database inspect each table inspect the data on each

[01:00:36] table and inspect the columns I made a I have a fix bug function because you literally

[01:00:43] pass it a line of code you pass it an error a PHP error it gets I actually gets triggered automatically

[01:00:49] so when my PHP error happens it triggers the fix bug function yeah and then it commits a pull request

[01:00:56] and I have another agent called jim and jim checks the code from bold which is the agent.code

[01:01:03] and then he outputs the pull request or yeah and then I get more jt.py one but yeah yeah

[01:01:09] it can do many things you know that it works because yeah it works because uh it's specifically

[01:01:16] built for my code base I it wouldn't work with another code base and also my the code base is

[01:01:23] I basically that is built on a framework that I made from scratch yeah so it's vanilla PHP

[01:01:30] yeah I know people are gonna say I'm crazy I don't care uh the advantage of this it has many

[01:01:37] advantages one of them that I never would have foreseen is that it's super easy the longer the

[01:01:47] more references you have you know the more complex the framework the harder it is for AI to code

[01:01:53] obviously because you have whatever you have jump around places or main levels maybe so my code

[01:01:58] base is not like that I have a function and I try to make keep each function without any calls to

[01:02:04] other functions so if I if I can avoid making the function in a functional function I avoid that and

[01:02:11] it makes it super easy then you can basically copy paste the function into jgpd or whatever and tell

[01:02:16] it add this feature or whatever change it and it works like 95 percent of the time

[01:02:23] so it's super easy to code with AI on certain code bases which is something people would never

[01:02:28] have foreseen when they designed the frameworks that we use today so the guy that built

[01:02:32] Marvel did not build it with AI agents in mind he built it with the constraints of human thought

[01:02:40] but now we have AI agents so if you have a favorite that's optimized for AI agents you're gonna

[01:02:47] spoil everyone else that make a lot of sense it's like it's very good idea so like if you're

[01:02:53] listening and you're searching for an idea build a framework made for AI agents not for human

[01:02:58] go yeah definitely I'm like because I have this problem you know like you exactly what you said

[01:03:04] I have this problem my code is split in different files each time I want to fucking ask the AI to

[01:03:09] do something I have to copy paste on every file every function to put it in one file to give it

[01:03:15] to the AI then otherwise he doesn't understand it's like yeah I lack context okay invent things

[01:03:20] right now I have one page just one huge page that's the admin and it has like I don't know

[01:03:28] how many hundreds of features you can do everything there and it's one page it's one file

[01:03:34] I don't know how many lines probably like two thousand lines and right now so I

[01:03:40] so these two agents that they have Bob and Jim they have their own gmail accounts they have

[01:03:45] their own linear account linear is like a project management they have their own GitHub

[01:03:49] accounts they have everything like I'm paying for their subscription okay so I can create I can create

[01:03:57] right now a task to be a feature a new feature for the admin so I created the linear I open it up

[01:04:03] create it assign it to Bob set the tab to admin and then he will call the feature and I even

[01:04:10] I even have demos where I did it live one shot he got the feature right in one shot with the API

[01:04:16] he builds the API everything everything and it works in one shot because it's one file

[01:04:23] yeah yeah he has all the context he needs yeah and now now with the cloud and everything you

[01:04:30] have a lot of context windows so it's super easy to before it was hard to give like enough

[01:04:36] context but now you know if you have like easy access to the to the even even fine even with smaller

[01:04:43] context you can use a trick which is you basically ask it to keep basically take notes basically

[01:04:53] as as far as every iteration yeah and like you ask it to build its own final prompt

[01:05:01] so it divides the big task into small subset of tasks each task when it finishes the result is added

[01:05:09] to the master prompt whatever and then at yet you run the master prompt this is how you do agents

[01:05:19] so you have a IA which is like a project manager creating tasks in linear then the other IA

[01:05:25] watching the tasks linear and do the feature the only thing the only place where I haven't integrated

[01:05:33] there is lady D'Antoni you have seven people in your company but it's just just seven IA

[01:05:41] there's a there's a clear rule so only only bots are allowed to have three letter names so they

[01:05:48] are bob jim and whatever whoever else comes that's a non-human he's gonna have a three

[01:05:53] letter name and humans are not allowed to have three letter names I'm only hiring four letters

[01:05:58] or more than two people who are gonna check your your recruitment because if you check our team

[01:06:08] or LinkedIn and you see someone with a three letter name that's not a human okay that's interesting

[01:06:15] why they have also LinkedIn account no no no I don't want I don't want anyone to post them

[01:06:24] imagine at one point you reached the moment where someone recruited you IA

[01:06:28] I mean uh yeah uh you know Elon Musk wants people to rent out their Tesla's you know robot

[01:06:37] taxifree something like that could happen you know with robots and agents with agents it could

[01:06:41] totally happen because you can you can run unlimited copies of the same agent at the same

[01:06:46] you can run you can scare you I saw I saw a startup doing that they uh it's like they

[01:06:55] they make an agent for you for 30 or months or something and you can you can run them yeah

[01:07:02] okay marketing agent or like designer agent or developer agent they are like four or five

[01:07:07] type different I don't know how good it is but I saw it this this is starting at one point

[01:07:12] like when you when you have like when IA will be good enough like paying 30 or months compared to a

[01:07:18] salary will be like maybe I just get one person which is a IA manager and that's it yeah yeah yeah

[01:07:27] okay shameless plug so if you go on tap there's a section called job impact index

[01:07:35] and it's a list of all jobs human jobs and the impact of AI on each job calculated by us

[01:07:43] and for each job you have a list of tasks for that job that are doable with AI and then for

[01:07:51] each task you have a list of tools for that task so it maps from job to tasks to tools

[01:07:57] and one of the jobs that has the highest that's most impacted by AI according to this calculation

[01:08:04] that we run is CEO so chief executive officer and if you look at if you if you open it up and

[01:08:12] look at the tasks that are impacted it kind of makes sense uh yeah yeah and I read somewhere

[01:08:20] where yeah I read somewhere about the Japanese game company I'm not sure if this was real news or not

[01:08:28] but they appointed an AI as the CEO of the company and it was running the company and everyone

[01:08:34] was super happy because they were they were saying things like yeah you know I can count that

[01:08:39] only that it's objective you know whenever it makes decisions it doesn't make it based on

[01:08:43] you know company policies and all this bullshit yeah so that's great you know I think if we

[01:08:50] have like models or pages that are advancing up maybe in five years maybe in 10 years I can see why

[01:08:57] a lot of people would want an AI manager you know because they're not yeah you don't feel like being

[01:09:02] judged by a human yeah it kind of sucks while being judged by a machine it's fine you know

[01:09:09] yeah definitely we're gonna have for me one of the main usage of GPT-4 when it was like

[01:09:17] with the whisper when you call was like I was having a I did a breakup and I was not feeling great

[01:09:24] and I didn't feel talking to anyone about that but I was able to talk to AI and I like process my

[01:09:31] brain breakup while talking to AI in January and there was yeah I spent like three or four hours

[01:09:36] talking with it and I was like that's fucking good that's super good did it forget what it

[01:09:43] was talking about after like an hour no because yeah I don't know it didn't it didn't forget about it

[01:09:53] probably because it had enough context you know in the yeah on the version 4 yeah yeah but yeah that

[01:10:01] was yeah that was super good okay do you do like a lot of voice chats with yeah yeah I

[01:10:12] like basically when I use charge epts by voice and the rest of the time when I use text is

[01:10:17] clode I don't use charge ept if I cannot talk to it I mean yeah the for me the best interface as

[01:10:26] voice is charge ept for now on your mobile so I use that and most of the time computer I use only

[01:10:34] on the truth there is one better interface but you gotta use a device that's not very comfortable

[01:10:42] to use which is the apple vision pro so I have an apple vision pro and the way I use charge ept

[01:10:47] I learned this from a friend super smart I think so you keep it you can put windows wherever you

[01:10:54] want so I have I usually working on my Mac so I have my like Mac screen then I have a separate

[01:11:01] browser with charge ept in voice mode but not voice chat where you can send it voice

[01:11:08] notes whatever yeah yeah and then we have a window with charge ept above where it's like

[01:11:14] god you don't look at it you don't see it and it's in the voice mode so that it's always listening

[01:11:21] like when I yeah so I get an idea or something I want to ask I just ask it aloud and somewhere

[01:11:26] response oh that's interesting yeah like you have someone in your room with you all the time

[01:11:32] that's listen I mean okay sounds creepy but it's very useful

[01:11:38] yes the only problem is that like after two hours of using that thing I haven't by the way I haven't

[01:11:43] used my keyboard like one because it's not comfortable at all and uh yeah everyone else

[01:11:49] I'm hearing from help us see yes you have the the the headband which is like more comfortable

[01:11:57] I I have both uh yeah I don't like it one I think the simple one is better but yeah anyway and also

[01:12:04] there are many apps there are I think it's probably not as successful would have expected a

[01:12:09] little bit more for the price but you know it's a nice it has a lot of things that are

[01:12:16] mind-blowing and it looks like it might be the future I'm not sure because it looks like uh like the

[01:12:23] Ray-Ban like from a this might be like similar to going back to what we said before of two paths

[01:12:30] one is uh you think about the perfect product uh you know that has all the features where no one

[01:12:37] has used anything similar before and this is like the Apple Vision Pro right invented a bunch of

[01:12:43] means a bunch of new concepts that weren't done before it has this thing that looks like is

[01:12:47] transparent but it's not it's actually projecting your face uh it has a lot of like insane technology

[01:12:54] it's an amazing device and then you have the uh metal Ray-Bans sunglasses that look like regular

[01:13:02] Ray-Bans and guess what I would rather wear Ray-Bans than a huge metal box on my eyes you know

[01:13:12] and it looks like that might be you know winning at least some race uh I can totally see I could never

[01:13:19] see someone walking around with a vision pro on their face like at least they're like making a

[01:13:24] YouTube video or something you cannot do that yeah I don't know but we'll see because like Apple has

[01:13:31] the power to educate people if they want so we'll see if they manage to uh yeah well

[01:13:37] I mean the only way is the lightweight version that's actually comfortable to hold on

[01:13:44] on your face I mean there was a there was a v1 so or v0 even let's say so let's see what will be the

[01:13:52] future I think for sure things will get like tinier and tinier and it will be like easier and

[01:13:57] lighter to have it but yeah we'll see of course and it's VR is not AR right I mean it's

[01:14:06] yeah it's both it's VR in terms of form factor because it's huge but it's AR in terms of features

[01:14:13] that it has transparency so yeah but it's not really yeah it's different kind of transparency

[01:14:21] because like for example HoloLens from Microsoft was here where it's like you see through and in

[01:14:27] top of that they put the element where the apple one is you don't see through but

[01:14:32] don't actually yeah they put camera that you can so it's a bit reverse and I think easier for them

[01:14:39] by the way that's really good it works really well like the tracking of your movement because it's

[01:14:46] as you said it's filming using cameras or the thing and showing you what the camera

[01:14:51] sees so there's obviously a lag and it's also processing things like you know light ISO and

[01:14:57] so on but it's super fast like you can move your hand like this and it's fine you can play games

[01:15:02] you can play you know people play sports just to show that works yeah and it's insane anyway I tested

[01:15:10] I tested the meta quest which is also having the same feature of pass through and I was so

[01:15:16] impressed like you see like no difference or I mean I think the meta quest is in black and

[01:15:22] white so you see in black and white when you see oxide world but still it's very impressive

[01:15:26] then you see in black and white I think in certain it's certain mode I think you can see totally

[01:15:35] transparent I'm not sure I mean with color but when you play a game for example if you get out of the

[01:15:42] box you've decided you see then he show you the outside world in black and white to don't disturb

[01:15:48] you from the game yeah yeah makes sense back to AI man yeah I think we can go to my finishing

[01:15:56] question and we had a good good good talk about meaning things there are one things I am I think

[01:16:03] I would like to ask now because it's a it's a good question it's like what's your for you the

[01:16:07] biggest success you had in the in your indie maker journey well it's definitely tough I mean

[01:16:14] nothing comes close yeah yeah I mean I mean not generally tough but maybe in tough while doing

[01:16:20] tough what was the big success you had like what something you did was working very well compared

[01:16:26] to the rest interesting question I think it was the first to have AI search I think that was a

[01:16:35] big one obviously I'm sure I have a gator and you don't have a search yeah but you know I haven't looked

[01:16:45] at it in a while so last time I checked the GPT store not saying it's an AI like nothing it's

[01:16:51] a similar to tough but whatever you would expect it you would expect that search to be AI and it's

[01:16:57] not so it's very weird but yeah that was one of them but I think it was many decisions one of

[01:17:05] them is tracking a lot of things and because I was alone for a long time running this the only developer

[01:17:15] doing everything myself I couldn't you know I have to make sure that I could at least sleep at night

[01:17:21] so how do you do that I would carry my laptop with me everywhere in the car I stopped many times

[01:17:27] on the side of the road because the server crashed because people were scraping it

[01:17:33] so but obviously you've got to sleep at some point so what I did is I set up alerts to monitor

[01:17:41] for errors or you know stats that shouldn't be things that shouldn't happen let's say and

[01:17:48] by the way I recommend this if you I'm sure that no one has heard the pot it's called it's a map

[01:17:54] called pushover it's also a website I use it yeah man super nice I cannot recommend it enough it's

[01:18:03] five dollars lifetime so just go buy it and then figure out what it does so what it does is

[01:18:10] it's a programmatic way to send yourself push notifications it has a native a native app

[01:18:16] and the app the best part is that the app has support for critical notifications I'm not sure

[01:18:24] how they're called in iOS but our notifications that bypass everything mode and everything else yeah

[01:18:30] so it's amazing and it has no limits like you can also use it for free and I use it for everything

[01:18:38] so I send whatever you know in my code I get somewhere where I shouldn't get I shouldn't

[01:18:45] get there I send myself an education and then I see it and I fix it immediately so that has been

[01:18:50] by far the single best decision I've ever made and I'm monitoring everything I'm monitoring

[01:18:55] crone jobs all the crone jobs all the errors all the situations where we shouldn't get all the

[01:19:01] sanity checks everything and it gives you peace of mind huge peace of mind yeah so I couldn't

[01:19:07] recommend this enough for specifically for indie makers go use pushover it's gonna be the best

[01:19:14] purchase you can do for five dollars lifetime yeah okay I have no affiliation with pushover I'm not

[01:19:22] but I just love it no I agree it's very good stuff and do you have only custom process for

[01:19:28] checking the errors or do you use also sentry or something like that no I wrote my own ph

[01:19:34] parser that parses the php blog error log and sends notifications for like critical errors

[01:19:40] like that okay and I also monitor automatically server hand stats also like a 50 line php code

[01:19:49] we can just check cpu percentage ram stuff like that and send the notification if they go above

[01:19:53] the threshold then check that every minute super nice yeah yeah I find a lot of resemblance between

[01:20:00] you and peter yeah I think he did a lot of things like that too yeah I actually spoke to him

[01:20:07] a few times I have I had no I still have actually I've been having his banner his twitter banner

[01:20:17] he has banner with dog that says you can do the team believe in yourself super cute dog so I have

[01:20:25] that on my wall in my bedroom and I've had that for years yeah I have like a mood board of I have

[01:20:31] two images so one is that and another one is an old couple sitting down on a bench and they're

[01:20:38] obviously mad at each other and the man is holding an umbrella above the woman's head

[01:20:47] and it says just because you're we're mad at each other doesn't mean

[01:20:53] just because you're mad at someone doesn't mean you stop loving them that's from my girlfriend

[01:20:57] she put that on and I put the you can do the team team okay that's cool and it's it's super

[01:21:04] motivating the the dog one is amazing when I thought it was never Zion's dog but it turns out

[01:21:10] it's it's not yeah he doesn't have he doesn't have dogs he talked about the dog's mess not so long

[01:21:17] ago I have I have question about like you're John and like we've talked about the thing like

[01:21:24] how you build your tools and the things but why you never went for example to VC money or building

[01:21:31] start-up the whole way that's a great question yeah look I actually so one of the things I did

[01:21:39] before theft like when I tried games I tried everything is I applied to YC I applied to Y

[01:21:45] yeah so I have a little sass called wait for it slide and subscribe I know

[01:21:58] okay this was before behind and so it was an embeddable subscribe form where you could type

[01:22:08] your email and then slide and subscribe and it will keep you authenticated so all over the

[01:22:11] web once you do it once you never have to read again that's pretty nice action technology and also

[01:22:17] I have I still have the domain subscribe not to subscribe to so paired yeah paired with that as a

[01:22:25] like the link tree alternative but it was like huge galaxy grain thing that I worked I was waking up

[01:22:30] at 4 a.m so that I could work on it before going to work and then I would work on it until

[01:22:35] like 11 p.m so I would barely sleep and I over engineered it I have like five different sign-up

[01:22:43] flows based on who you were if you were a subscriber versus if you are a website owner

[01:22:48] versus if you are a medium owner if you were a mobile device I made 79 dollars yeah that's how

[01:23:06] I worked on for like then I completely burned out for a while anyway so with this I applied to YC

[01:23:13] I applied to Y combinator and when I didn't know you know what what it takes I just thought

[01:23:19] what when I opened the application have you ever filled in or looked at the application for

[01:23:26] I think I watched but never never did directly okay so one of the first things they ask you

[01:23:33] is first of all they ask you for a co-founder they say well you you must have a co-founder

[01:23:39] basically we they almost never accept single founders which is I think it's stupid but not

[01:23:44] for them I mean it's stupid for the founder I don't recommend I don't get a co-founder unless

[01:23:50] you absolutely need it and then they ask you why is this a billion dollar company why is this

[01:23:59] hundred billion dollar market or what like why is what if I'm fine with a 100 million

[01:24:05] dollar company or a 10 million dollar company I find with a 50 million dollar company

[01:24:10] they ask you a bunch of it's because of uh misaligned incentives basically between

[01:24:19] they have a set of incentives and you have a different set of incentives

[01:24:26] and yeah so they also ask you to like move to Silicon Valley stuff like that I'm not sure if

[01:24:31] they still do but they it was heavily encouraged yeah after covid I think they changed the rule

[01:24:36] no I think they changed it yeah yeah but anyway I'm not seeing why he is bad I think it's the

[01:24:41] best like startup school on the planet but I think they I disagree with a lot of

[01:24:49] uh let's say they're teaching or whatever when it comes to like small startups tech startups

[01:24:58] things that you can start on your own like I think you can you can uh you should

[01:25:03] before raising money you should probably make like 10 000 a month just to show it works if

[01:25:09] you don't especially in AI because right now if if you have an idea in AI first of all you

[01:25:14] can if it's a simple website or whatever you can build it yourself uh and if you don't have traction

[01:25:20] of course you can fix it but you're better just make up something else because like a third of ideas

[01:25:27] get traction you just change the idea it's faster look with theft I got 100 000 visitors in the

[01:25:34] first week and I'm seeing like you can see like viral AI all the time every week

[01:25:42] so if you don't have traction just get another one plus

[01:25:45] you know a million other problems like for one other thing I don't agree with from the general

[01:25:53] wisdom is like you've got to fix retention super fast I think you should you're better off

[01:26:00] optimizing for a total number of monthly active reserves or whatever frequency

[01:26:06] works best for your weekly daily active reserves and so on and then you can fix retention later

[01:26:11] you actually sometimes you have a better chance of fixing retention later once you have more users

[01:26:16] once you have network defects yeah plus plus early on if you focus on fixing retention early on

[01:26:24] what happens if you need to pivot you wasted a bunch of time optimizing

[01:26:31] optimizing shit yeah so it's like okay let me make sure that at least this is gonna be something

[01:26:38] that is correct have the right assumptions that we proved that we have we have the data from users

[01:26:44] showing that you know we were right or we're on the right path let us pivot if there's a need and

[01:26:51] oftentimes I think there's a need to pivot at least slightly you know you don't need to pivot

[01:26:56] completely like probably change the niche completely but people slightly to you usually discover on

[01:27:03] ad yes and market or an ad yes and idea that's like super small something someone mentions or a weird

[01:27:11] way in which someone uses your app this is usually uh this is how you know many many apps

[01:27:19] work um so yeah yeah I think that's why I don't really see plus uh you know there's the money thing

[01:27:26] the money thing I mean if you want to make money you're better off working on a smaller company

[01:27:33] you don't need to build a billion dollar company first of all it's much easier to build a 100

[01:27:37] billion dollar company than a build a 1 billion or a 10 million dollar company

[01:27:42] way easier than a billion dollar company plus there's this thing especially with AI I don't know

[01:27:47] I think it's extremely dangerous for AI founders if you have a difference in a basically multiplier

[01:27:55] between your valuation your startup valuation and the market value of your startup and by the way

[01:28:03] these are different things so the valuation and the market value so market value is how much is

[01:28:08] your startup worth if you were to sell it today on the market how much money would you get

[01:28:12] so let's say you raise 20 doesn't matter at what whatever valuation right you raise 20 at

[01:28:21] a 100 million dollar valuation okay and your value your market value is less than 20 of that

[01:28:29] valuation so your market value is less than 20 million your market value is 10 million 50

[01:28:33] million whatever because of something called liquidation preferences if if you are forced

[01:28:39] to liquidate or if you whatever if you liquidate for whatever reason you get zero yeah so however

[01:28:47] much you you gave away to investors let's say you gave away whatever 20 percent of 50 you gotta make

[01:28:55] sure that the multiplier between the real value of your company and the valuation is less than

[01:29:01] a fifth less than five you know because if it's more the difference is more than five x you're

[01:29:07] which is something that happens a lot turns out a lot of founders you know

[01:29:11] sell their companies for millions and make nothing more you know there are you know there are ways

[01:29:16] to defend yourself from this you can have contingencies you can have you can negotiate

[01:29:20] like a small amount that is still it still sucks you know you cannot wait for that

[01:29:26] and come on if you yeah for me the crazy part in this is that has been so much

[01:29:33] the only way of building for so long where it doesn't make much sense like you know like

[01:29:40] for me going to vc is like you have a big ego and you want to do like something like

[01:29:47] big and people will remember but not everyone is like that there are many people that they call

[01:29:52] i call them artisan they like to do something that's matter to people and they don't need

[01:29:58] to be the biggest in the world they just need to have an impact and make good money from that

[01:30:03] and that's what i am for example i don't want i want to be an artisan i want to have like my tiny

[01:30:08] shop so which is my starter don't have a lot of interaction with people building the way i want

[01:30:14] not have to report to anyone this is for me what is important is my freedom to build the things

[01:30:20] i want and if you build like something with vc you will not have your freedom because you

[01:30:26] have people to report to you have to go the direction they want and do things

[01:30:31] so it gets way harder but you will have your ego feed because you can be media at one point you can

[01:30:37] be like you know whatever dinner with important people if you manage to grow the direction you

[01:30:44] and if you manage to stay because this is also something people forgot is they can start a

[01:30:49] startup but they are often not the right person to keep it growing till a certain point

[01:30:53] so you get you get kicked out okay okay you just become an investor of your of your startup but

[01:31:00] you're not like the CEO anymore because you're not competent which is normal like you have come from

[01:31:05] zero to i don't know one million or ten million and then going to ten million to one hundred

[01:31:09] million is different game this usually happens if you raise money so if you grow by raising

[01:31:17] money that's a fake way to grow the value of your company i would say in my opinion the

[01:31:23] real way is to grow the value of your company through the the profit it generates

[01:31:27] but that's the purpose of the company is to make profit there is no other it's a company okay it's

[01:31:32] a for if it's a for-profit company of course if it's a non-profit company whatever if it's a

[01:31:37] for-profit the purpose and the success of the company is dictated by how much profit it makes

[01:31:44] so if you notice that for people who start from zero okay with no investment and there are

[01:31:49] some you know bootstrap successful company the founder is that he's actually the right person

[01:31:56] to run it until the end it doesn't happen that you know the guy who started the water and name

[01:32:01] whatever big company you want from scratch then uh you know he turned out he wasn't good enough

[01:32:06] to run no he was you know because turns out it's very hard to take a company from zero to even

[01:32:11] ten million a year you know it's extremely without without raising yes yeah yeah yeah i totally agree

[01:32:18] you and also i think what's changed is the time you're gonna go to one to 100 million

[01:32:22] by bootstrapping it's a bit longer that it's probably longer than raising money so you are

[01:32:29] now the best person knowing because you took the most time into it while when you raise money

[01:32:35] sometimes you go super fast in one two years so you don't have barely barely experience

[01:32:41] exactly plus this exactly amazing point this doesn't give you time to pivot

[01:32:48] oftentimes this is why in the bootstrap model you can postpone fixing retention as long as you

[01:32:54] have high enough growth and you can postpone fixing a lot of problems that you cannot postpone in the

[01:33:00] height where it's scaling the VC system plus you know there's there's another huge advantage

[01:33:06] that obviously everyone who bootstraps their company knows which is okay if you get tired

[01:33:13] you can just sell it and make a retire if you if you are successful enough you can sell it and then

[01:33:20] retire so it's amazing you cannot sell whenever you want uh if you're not the sole owner you know

[01:33:30] you have to talk with the other owners and yeah and and finally often you're not getting way much

[01:33:38] for like i had a friend he made a very good startup like million dollar company and when

[01:33:44] he sold it at the end he has like very few percent left for his own part of the company so when he

[01:33:50] sold it make him he make basically like less than a million on his for himself well for all this

[01:33:58] less than he would have earned as a seller he most likely working for a good company yeah and

[01:34:04] and you see other indie makers like they do a project for like six months eight months they

[01:34:08] sell it for 200k and you know like if they go faster or bigger for one two more years they will

[01:34:14] have sold for bigger and it will be way easier than doing like the VC way for for i don't know

[01:34:21] six seven years where it's like you kind of dying every day because of the amount of work

[01:34:26] depends on uh on your time discounting preference which is how much you value things right now

[01:34:32] versus in the future so for me for example it's it's around the an order of magnitude per decade

[01:34:39] so like uh for me it's roughly i'm roughly indifferent whether i get one million today

[01:34:44] or 10 million in 10 years or vice versa if uh if i have the opportunity to get one million

[01:34:51] today instead of 10 million in 30 years i'm okay with that so if you if you if you know what

[01:34:57] the value is for you and it's unique for each person and that pitch it can differ in time

[01:35:03] for each person uh you can calculate what the annual growth rate of your business has to be

[01:35:09] in order for you for for it to make sense to sell it or keep it so in my case because it's yeah

[01:35:16] because it's 10x per decade it works out to 25 percent per year for 10 years so if it grows

[01:35:25] less than 25 per year i sell it if it grows more i keep it okay that's super clear it's actually

[01:35:32] yeah it's super clear i actually like to take a like a margin so it's like 50 percent uh but if

[01:35:38] it grows by more as long as it grows by more than 50 percent a year it makes sense given my

[01:35:44] time discount preference to keep it if it grows less it makes me to sell it

[01:35:49] and i can sell it to someone who has a different time discount preference so it

[01:35:53] doesn't mean that just because i sell it i undersell it yeah for sure for sure that's a very good advice

[01:35:59] thanks for that uh i think we can go to my my finishing question um if you if you had the

[01:36:07] the chance to like i i will manage to do this sentence what do you what do oh no it's over

[01:36:16] it's over you see i see i i prepared food like before going to the meeting and i wanted to have time

[01:36:23] and he didn't have time and you see now i am like i'm lost yes i'm i'm lost is that a pizza over there

[01:36:30] yeah it's a pizza it's a pizza oh man i love pizza usually don't eat pizza much but i don't know

[01:36:36] yesterday i felt like oh i could have one yeah so but anyway so the question is uh what you will

[01:36:43] wish to know uh earlier in your journey before you started like someone told you something you

[01:36:49] find like he didn't know that was important i'm gonna go back to what i said about indie hacker

[01:36:55] mistakes i'm gonna i think one that i haven't mentioned is zero marketing budget which is

[01:37:05] a mistake the time limit test as well uh it's a temptation you know to because you get bragging

[01:37:11] rights right you can say yeah i launched it with zero marketing why would you launch it with zero

[01:37:15] marketing put 500 dollars into it no i'm not saying put a million dollars you know there's a huge

[01:37:23] difference between 500 dollars in marketing budget and zero marketing budget and i don't i don't

[01:37:30] take it smart as long as you believe in your idea you actually done your homework when you

[01:37:35] researched it and so on and we actually validated a little bit before launching uh then you should have

[01:37:42] a launch budget you know because uh you get you get way more traffic if you have some kind of budget

[01:37:48] and you get more data and get to learn faster so even if you don't have let's say a few hundred

[01:37:55] dollars uh to launch you can work on other things do whatever gig jobs set aside a few hundred

[01:38:03] dollars launch your idea so i think this is what i would have done and i think it would have made

[01:38:09] the difference for a lot of my failed projects when i always launch them yeah i'm gonna post it on

[01:38:14] product hunt and uh i'm gonna ask everyone i know to upload it and that's my launch strategy

[01:38:20] or and post it on reddit and the hacker news i think it's better to maybe spend a little

[01:38:26] bit more time during research and validation so you have higher conviction and when you have

[01:38:32] higher conviction you will usually want to invest a little bit into marketing so i think uh that's

[01:38:38] one of the mistakes that i won't be making again hopefully thanks thanks for that what you will

[01:38:45] put in the 500 euro budget if you like you were someone starting like where you will put the

[01:38:51] 500 euro uh listing on different places including past obviously if it's an ai at

[01:38:59] what it doesn't have to be uh okay it could be uh it depends a lot on what it is if it's something that

[01:39:05] works well on twitter uh could be twitter ads or could be actually much better than twitter ads

[01:39:11] you can buy a shout out from an influence so if it's something that works on tiktok just give

[01:39:16] 500 dollars or whatever it could be a thousand dollars but it's not gonna be 10,000 so let's

[01:39:21] say a thousand dollars maxi choose the best influencer and it's usually like a smaller one

[01:39:28] so look for influencers with under 10,000 followers so that they don't have an agency

[01:39:34] that you have to deal with uh you know they're they're still doing their own thing they're just

[01:39:38] one person uh and you can take time and there are websites that you can use to find the right

[01:39:45] influencer find the right influencer with influencers it's a bit more risky because

[01:39:49] if it doesn't work out in terms of conversions you don't get residual value which you get with

[01:39:56] listings so if you get listed on directories like that even if your launch doesn't succeed the

[01:40:03] financial terms uh you don't make money during that day you still get a lot of SEO value you

[01:40:11] you get distribution forever you're still there you're gonna get traffic for you know the next

[01:40:16] five years you don't get a lot of residual value with influencers this is why i don't like doing

[01:40:22] influencer marketing too much because it's the type of sales that you have to do

[01:40:26] it's one thing you know the yeah and you have you know the it's sales like the old type of sales

[01:40:31] that every month we start from zero of course you can if you do it at scale but then we're

[01:40:37] talking huge budgets it can grow your brand you know increase brand awareness increase

[01:40:44] brand searches then everything else have all kinds of secondary effects but not with one influence

[01:40:50] so that's my advice okay thanks for that do you have maybe a favorite quote before we finish

[01:40:59] uh no i'm not even something you repeat yourself often i like i like uh i have this

[01:41:09] which is from it says only the brave right from uh not gonna name the brand but it's a well known

[01:41:16] brand i like it yeah only the brave okay cool thanks for that um so now we have two last questions

[01:41:24] so one is for you one is for me super easy so uh for me first what's the next person i should

[01:41:31] invite after you in this podcast before he got exploded so maybe i should invite him again

[01:41:41] then invite mark again uh did you have did you have levels no not yet well then there you go

[01:41:50] if you have a contact i'm i'm willing to take it okay and so the next question is for you where

[01:42:00] should we should send the makers listening the podcast we want to know more about you

[01:42:07] uh you can follow me on twitter uh the handle is i make cool sites oh yeah

[01:42:14] yeah i didn't have any better ideas when i thought about this handle for like a bazillion years and

[01:42:21] screwed it up yeah oh yeah okay i will put it in the description this is all i had that's it i'm sorry

[01:42:31] you're better to take a website domain

[01:42:36] but it's it's nice handle in fact uh so thank you thank you for your time that was super nice

[01:42:41] to have you in the podcast in the show glad to be here all right good luck with the podcast

[01:42:48] thanks and for our listener if we have uh if you have enjoyed listening this podcast uh you can

[01:42:53] still send a message to me or to him to andre like to say thank you if you have any question

[01:42:58] any things is always nice to have like some comments of what we did check his website check

[01:43:04] there are a year for that or i don't know to talk anymore definitely i need food and see you next

[01:43:09] week for the next episode bye bye thank you cheers