I Forced Myself to Build $1M Business with AI

0:00So, in this video, I'm going to talk about how I built an AI app solo and took it from zero to $1 million in annual recurring revenue. And I'm going to talk about exactly how I validated the idea, got the idea, launched the idea, and all the things that went wrong during the launch. So, you can hopefully learn from me, and do it even faster for your own app. This was the very original video that I posted August 1st, 2024, I think, when I first got the idea to build an app that would repurpose your long- form content into short form content. And for me, my use case was

0:30just like I wrote my blog post once a week, my YouTube video once a week, and I just wanted to turn those into short form social media posts really quickly. So, all I did to validate the idea, since I already had a little bit of an existing audience at this point, was I just shared the idea, made a Tik Tok out of it, and asked people to comment if they resonated with any of the pain points I described in the video. And over 200 people commented. So my next step was then I DM'd the people who expressed interest and asked them like more questions about their pain point,

0:59what tools they use today, how much they're paying today, what type of content are they making, which platforms are they posting on, etc. So yeah, that was it. So my very first kind of like exploration of this idea was August 1st, 2024. Just really asking people if they wanted the tool really just had the idea, then posted the Tik Tok and then I DM' people to set up a few calls. like it was probably less than 20 calls in total before deciding, okay, I'm going to go build this. And that was it. That was like literally the start. Really not anything high pressure. Like a lot of

1:31people like are so worried about choosing the wrong idea and everything. But my mindset was always if this completely fails, like I'll just open source the code and give it away for free and make content out of it. Like that was my mindset at the time because I'm a creator. It didn't seem like a high-risk thing. It seemed like, okay, I'll just use this as an opportunity to experiment with vibe coding tools and see how far I can take it. Okay, so the next thing to do was trimming down the vision. So the vision was like you have all your social media platforms, AI powered remixing, you upload your brain,

2:00your brand voice, product info, services, AI will filter content for quality and relevance, repurpose that content using your brain and brand, publish anywhere, and then you can continuously teach AI with examples [music] you like. I thought of it like pandock for content creators. Like pandock is a thing that helps you convert media and data from one format to another. So that's how you thought of it uh in a more technical way. But this was like I think late August or early September 2024 where I just shared like my vision, my spec. I had like a really basic user prototype here, like just literally a sketch and then I went

2:35through the video trimming stuff from my spec and from my prototype. So all these red boxes I'm drawing, that's because I'm trying to remove complexity from my initial version. And again, I filmed myself doing all of this. So it's all public record now. [laughter] Um, so basically my first version was that you like put in some text like let's say you wrote a blog post, you put it in and then you want to repurpose it to a Facebook post. Boom. That's it. And this was a seven video series using I think cursor AI at the time. Back then AI coding agents were very limited. Like even with cursor, it didn't have

3:07composer. I had to create and delete files manually. Towards the end of my series, they eventually released Composer so I could finally just have an AI agent create and delete files for me. But yeah, so this was just like V1 prototype that I built out. And if you kind of play later in that series, you'll see, you know, what it ended up looking like. So if I go all the way to like part Okay, so like this is basically what my MVP looked like. You create content and then uh you put in like some a long form piece of content right there. So like that's the original

3:38content that you are trying to repurpose over there. Okay. And then you choose the platforms you want to repurpose it to. In this case, Facebook and Twitter. You choose if you want an AI image generated along with a post and then it does it. And this was using like Dolly 2 at the time. Like it was terrible. [laughter] Like lots of typos in the images. Like we have it so good now. Like we forgot 2 years ago how terrible a lot of this stuff was. Um [laughter] so that was my V1. Okay. So really really ugly. really crappy, but I went ahead and launched it anyway. So, around

4:10October 1st, I opened my early adopter program, and it was basically like a newsletter I sent out to my existing audience saying, "Hey, fill out this Google form if you want to join my weight list." This was the weight list. Like, this is just a real screenshot of it. I think I have the real thing actually still live. So, this is the act. You can still sign up for the Blueot weight list. Apparently, this Google form is still active. So, I asked, you know, what platform do you post on? What platforms do you repurpose your content to? What type of content do you post? How do you post from your

4:40phone or your computer? What's your workflow today? And what's the most timeconsuming part of your workflow? So, I highly regret actually doing all of this because um what I realized is what people said did not match at all what they actually did. Like not even close. [laughter] Like someone would say, "Oh, my main platform is Instagram and I want to post the LinkedIn." And like when I actually see what they're doing, their main platform is Tik Tok and they don't repurpose content anywhere else. So yeah, people's answers did not match at all. So the early group of around 50 early adopters I had just were they just

5:12had completely different use cases and honestly it left me more confused and frustrated. So I actually regret doing this early adopter program and I wish I just launched publicly earlier. Looking back, part of the reason of doing the early adopter program is cuz like I was a little embarrassed by how like ugly and disgusting my MVP was. Like you couldn't even publish content um not in the V1. Like it would help you write the content for, you know, other platforms, but you couldn't actually even publish it. Uh which is like the main thing my app does [laughter] today. But yeah, you know, I look back on this. So just be

5:45careful if you are creating an early adopter program. like if you're not super super strict about screening your users and the pain points they have, you may be pulled in like a hundred different directions. And that's kind of what happened here in this early adopter program. Just imagine like 50 different people with totally different use cases, different platforms they're posting on and like different things that they want. And that made it really tough for

6:07me. But then I was like, okay, whatever. This isn't working. I'm just going to launch it in public. Whatever. [laughter] That was literally my thought process. So I figured well you know maybe out of thousands of people who sign up there will be a handful core group of people who actually find it valuable and then I can work from there. Um and that is exactly what happened like a lot of people signed up but really only a handful stuck around using it on a consistent basis and those are the people like I watched regularly to see what they were doing and just built around their use case. The launch is not

6:38anything fancy. So let's just talk about what I mean by launch. So, my launch day had a lot of failures, by the way. We'll talk about that. But it was essentially like like a 5minute YouTube video demoing the product. Uh, one newsletter announcing like, "Hey, this is my product. Come sign up for it." I think I did one Tik Tok video and I think at I'm not even sure at the time if I had Instagram or Facebook. Like, I don't think I did. So, and then probably like a LinkedIn post because I was on LinkedIn at the time, but yeah, I wasn't

7:09actually on many platforms at the time. Um, but that's it. That was all I did for my launch. Like there was no press, there was no PR, there was nothing, no fancy launch party. Honestly, like a typical day making content. Just this time the content would be about my product, why I'm building it, and and who it's for. Um, but that's it. So,

7:27like, don't over complicate your launch. Um, I want people to understand like no one cares. Literally, no one remembers your launch. Like, do you remember the day Google launched? No. Do you remember the day Facebook launched? Like, no. Like, do you remember the day Snapchat launched? No. Like, the launch is not a memorable day for like anyone except you, the founder. So, like, don't stress over it. Just get it over with as fast

7:49as possible. That's my honest advice. So, here's a screenshot of part of my launch email where I'm explaining, okay, here's why I'm building Blueotato. Growing on social media takes so much time. It's overwhelming. If you feel this way, you're not alone. I've heard this painoint again and again. I've experienced it myself. Blah blah blah. These screenshots, I put a lot of receipts. Like these screenshots were from the last 7 days and it's 100% free organic growth. And so that launch day was October 20th, 2024. So I sent the email, posted the video, opened the doors, and then basically everything broke. So [laughter] yeah, let's talk about that. Um, I even

8:22made content about it cuz that's what I do. So here's a video where I'm just like recapping all the things that went wrong with my launch. So the biggest one was um I was like super nervous with a YouTube video. So like I I messed something up. But I think I filmed like a different version of the YouTube video and then I put the wrong link in the newsletter. So when thousands of people opened the newsletter, the YouTube link was broken. It was just literally a frowny face. [laughter] Gray box frowny face. That's what they saw. This was literally screenshotted from my newsletter. Like blah blah blah

8:52so excited about everything and like this is what you see in your [laughter] your email inbox. Obviously, a lot of people, you know, emailed me back saying, "Um, I think your link is broken." [laughter and gasps] And I had to like reply to everybody like, "Oh my god, here's the correct link." Uh, etc., etc. But yeah, that was a pretty big launch day failure. I mean, like, if I was heading marketing at a big company and I did that, dude, I would probably be fired for that. [laughter] But since it's my company and I could

9:22just do whatever I want, it it was fine. Like, now it's a story. I made content out of it. Okay. And yeah, that was not even my only failure. So, I'll I'll list a couple here. Besides that, let's just make a list of like launch day failures. Yeah, main link in the email was broken. Number two, it was not mobile responsive. So, people visiting my website on their phone. It just looked like a total mess. Um, a bunch of signup payment methods were not working. For example, like Apple Pay, Amazon Pay, and then I I think I offered to like book meetings with people if you had signed

9:55up. And then those calendar invites were missing meeting links. And that's just like four of them. There were like a couple more I recap there. People get so stressed about launching. And I don't know. I I just feel like stuff will always go wrong and you just have to get over the launch. Like I said, like no

10:10one is going to remember your launch. It's really just a big deal to you. Everyone else is just like, "Well, can I buy it? And does it work?" Okay, [laughter] if it doesn't work, I'm not going to pay for it. If it works, I'll continue paying for it. Like that's very simple. [laughter] For me, the most important thing that it it improved, people were like trying to click the link, trying to watch the demo, they were trying to sign up, and they were trying to buy. So, all I had to do for the next couple of weeks was just fix everything that was broken. You know, like fix Amazon Pay or whatever

10:39wasn't working on the payment side, fix mobile responsiveness, fix the YouTube video link that didn't work. You know, I just had to like go fix those things and there were already a stream of people wanting to come in and check it out. Um, and that's the value, by the way, of building a personal brand. I know I repeat this a lot and you hear this from a lot of other people. But it's for

10:57startup founders, it's like cheating. Like I built a startup before a personal brand. That was like really hard and depressing [laughter] with a personal brand. Like despite all the mistakes I made, okay, despite all of those like really junior mistakes, I still had a flow of people coming into my funnel, trying the product, and a small percentage of them actually willing to pay to stay, you know, despite all of those mistakes. There was always had like a constant stream of people coming in and checking out the product. Oh, yeah. So, a lot of people ask me like, okay, what about my website and stuff like that? So, let's just talk about

11:31just really basic stack. So, for my MVP that was built entirely with cursor AI. My website was a $60 web flow template. Like, I didn't even design it. It was like $60. And I just changed the text and the pictures and that was it. And I had not updated this website for like 1.5 years. So, I had intercom as my support widget. Okay? So, like if somebody had questions, they could type

11:54in the chat and eventually I'd reply. Um, and then I used Post Hog to watch user sessions. And this is honestly like the main thing you do when you initially launch a product. And so a lot of my initial work was just like literally sitting there watching this. It's like Netflix. Like I'm just binging watching all these user session recordings, seeing what people are doing and trying to remove friction throughout the process. So if this is step one and the onboarding, step two, step three, my goal is to just continue improving the product to remove friction so that more people get through each step. That was

12:25the only goal I had in the first month or two in terms of product development. So, about 1 week after launch, I posted about 10 things I learned launching my AI app. Number one was launch faster. I I flip-flop on this because like I guess I didn't treat Blot that seriously. Like I said in the beginning, I viewed it as low risk because even if it completely flopped, I could just give it away for free and my community would be like, "Yay." So, I didn't really view it as a serious thing. Number two, I spent way too much time on the early adopter program. Uh, again, that that was very

12:58confusing to me and like I wish I hadn't wasted my time with that on. I wish I just skipped it. I had onetoone onboarding calls. That was a ton of work, but I quickly learned what to improve. Um, I think I priced it at $49 per month. There's no like real reason. I just looked at some other competitors and I was like, "Okay, let's try this out." [laughter] Um, I eventually dropped the price just cuz like, you know, I feel bad for people international and like business owners

13:21that are just starting out. Oh, yeah. So, let's talk about vibe coding. So, I get a lot of questions about this. So, like, can AI really build a production ready app? I mean, I used it to build version one, so I would say this was like 100% AI. Uh, I really did not look at the code base, honestly. Like, it was a disaster. Like, the code base was terrible spaghetti code. Keep in mind, this was like late 2024. AI vibe coding tools were just not very good. But, I think it's normal. Like, I think it's okay to use AI to completely vibe code your first version. And then as you

13:53actually gain more users then you want to switch to AI assisted coding. So what I mean by this is you still have a human ideally a senior developer driving AI and what it's doing. And it's normal to switch this way. Like I tell people the goal of vibe coding is to validate your product get paying customers and then move on to the next level with high conviction. Hire a developer to take everything over and like build a real product. In my case, I didn't hire a developer because uh my first startup was like really hard like building uh training like speech recognition models and natural language processing models

14:27to work in real time low latency uh really noisy environments. Like that was really hard. This is an API wrapper. [laughter] I didn't think it'd be that hard. So, um a little bit of arrogance. So, uh yeah, I just didn't hire a developer for like the longest time. But anyway, oh yeah, my eventual stack in case people are wondering. So, like Nex.js, JS, Fastify, Superbase, and Heroku. And then for my website nowadays, I redesigned it recently with um Claude design. So I'll say Claude,

14:56and then it's Astro deployed on Versell. Uh just cuz my web flow ended up costing like $600 a month. Um and then my original intercom support agent, I moved that to NAND and that also saved me about $1,300 a month. Okay, but that's like basically my stack today. Uh but yeah, the key thing here is like, you know, I I still think vibe coding is amazing. Like it gives building blocks to people who could not build this stuff before. Like you don't have to pay $5,000 just to test an idea. Okay. So use vibe coding to test many ideas very quickly. But then once you figure out

15:28which one you're going to double down on, then hire a real developer to help you like take it to the next level. That's pretty much it. I guess I'll just conclude with with like revenue growth and trajectory. So, my revenue chart basically looks like this. I'll try to draw it. It's like a line. It's like literally what my growth chart looks like. Um, I wish it was, you know, a beautiful exponential curve like the stories you hear about with product market fit and rapid insane growth, but it's really not. Like, it's pretty linear at the end of the day. And these plateaued sections here and here, like,

16:01no joke, this is ski season. [laughter and gasps] Like I checked cuz you know the first 244 season and the 2025 season. I was like, "Huh, my revenue is plateauing right around late December all the way through January. What's happening?" Oh, I'm skiing. Okay. [laughter] Uh I'll try to fix that next year. Now I know it's ski season. I'm going to like try harder. But yeah, I mean this is my growth. It's kind of it's in a way frustrating. Like it's it's not exponential. It's like linear. Some months are slower than others. Some months are a little bit faster than others. Not a huge amount of variance

16:33though. There's no spikiness. There's no like sometimes you hear about like crazy viral trends and then they drop down like this. But yeah, no, not here. Just boring linear growth with plateaus during ski season. That's like literally what my growth [laughter] curve. So other comments I had is like, yeah, like I personally don't feel I have product

16:53market fit in the Silicon Valley sense. It still feels like a lot of me pushing make sure everybody knows about this, make tutorials on it, etc. But at the end of the day, like it is insane what you can build with AI solo or with a tiny team. But yeah, it's like pretty incredible what you can do. However, the main thing I want to emphasize is like the sequence of things that I did. It wasn't like build product and then get an audience. Like I actually had an audience first, used content to test the idea and test demand for it. So number one content to test demand for that

17:25particular product. Okay. And then like I said, I posted that Tik Tok video August 1st and it seemed promising. Then I used AI to vibe code V1 and it was really embarrassing. Like that's a pretty embarrassing V1. It is ugly. [laughter] [gasps] And then you know because I had the V1 now I could get users and like see what they're doing. Don't listen to them. No offense [laughter] to all of you amazing Blot users, but you know often times what people say is not what they do. So but once you have users you can

17:56literally watch them using your product. And then my official launch was October 20th, 2024. And it was just content. Like there was nothing fancy [snorts] about the launch inherently. No press release, no fancy PR firms. Like TechCrunch didn't pick it up. Like I didn't launch on Product Hunt. I didn't even launch on Twitter. It I just like kind of launched it to a community I already had in the channels I was already making content. And then revenue. Um I had I got revenue even from the initial launch. I think I hit 10K monthly recurring revenue within 10 days. I don't normally share like specifics of numbers, but it's so far

18:30away now that I'm comfortable sharing that. But yeah, the revenue like once you get revenue in, then you have conviction to keep going. Um, in my case, I essentially had to rebuild the product. Like I said, the first version from Cursor was vibe coded spaghetti code like that. There's no way like I'm going to be able to maintain that or easily extend that. But getting to 10kmr gave me the conviction like okay I should probably like treat this seriously now and like try to rebuild it for real like a real codebase. That means like actually planning the architecture understanding tradeoffs choosing good variable names writing tests writing comments writing

19:05documentation like all of the good engineering practices I kind of just skipped in my [laughter] V1. So, uh, yeah, that was pretty much the process and this was the order. And it was very chaotic. Like lots and lots of things went wrong that I didn't even share today. I just shared some of the biggest bloopers. But this is the process. This is the work. If it feels chaotic, like don't worry about that. It is part of

19:27the learning.