How to Build an AI Business from Scratch

Over the past few months, I've been talking to a lot of people building with AI, which is super exciting right now, right? Founders, investors, CEOs, and asked them one question. If you had to start from zero in 2026 with no code, no team, no funding, just AI and a laptop, what would you build >> [music] >> to make your first $10,000 a month? And here's what I noticed. We keep reading headlines about someone raising a hundred million for their idea or reaching crazy valuations, and we look at that and think, "Oh, that's what success looks like. That's the goal." But the thing is, there are many more people quietly making real money somewhere else entirely. I talked to Daniel Priestley, a UK entrepreneur of the year, who took four of his agencies, automated each one using an AI platform. And one of them, Score App, started as a $15,000 client project his team used to build by hand. It now has 8,500 customers, grows 4% month over month, largely bootstrapped, no crazy costs, just money. Nobody wrote a headline about that, and that's the actual opportunity right now. Take a small problem, solve it with AI, and make 10K a month. So, in this video, I pulled the exact playbook from people who've actually done it. What to build, how to find your niche, how to get customers with zero budget, and when you actually need investors. Spoiler, probably never. I keep hearing this from so many entrepreneurs that are like, "Just start making money. It's so much easier now." And I want to start with, of course, the legendary Gary Vee. Because when I asked him, "If everything crashed tomorrow, VaynerMedia gone, followers gone, what would you do with just a laptop and AI?" he answered this. If I lost my reputation and my personal brand, but I kept my knowledge of how to make content on social, I would build an app that's five to $50 a month, and I would make unlimited organic content on LinkedIn, Twitter, TikTok, and try to get customers. When you layer what I've been obsessed with about the last 20 years, which is there is a advertising and brand building ecosystem called social media that does not cost you money. And that's the point about AI apps. It really doesn't cost you money anymore to build something in the way that I grew up. Um but the fact that you can actually get awareness, that you're one TikTok away from having people know who you are and building a platform, that's that's intoxicating. Build an app, make content, get customers. But what if I'm not a developer? How do I actually build the product? Here's what the CEO and founder of Replit, Amjad Masad, told me. Everyone has an idea. But for the most part, the thing that's stopping them is that they don't have the technical skills or they don't have someone, you know, as a programmer growing up, all my friends were like, "Oh, hey, can Can you program this idea for me?" Well, now you could do it. We have a lot of stories where people have built their dream apps, and they've had these ideas for like 20 years. Mhm. We're talking about CFO at a VC firm. He's a domain expert. Like he knows how to manage a VC fund, and he never found the right tools for him. And he had all these ideas on how to build them, but you know, it's almost it's always hard to find engineering resources. So, he he used Replit, and in 3 months, he built his dream app, and he went out and sold it and got a lot of contracts, and I think he made He's on his track to make 5 million ARR, quit his job. Now he's an entrepreneur. Have you seen any big mistakes that people make when building something with Replit? Cuz I know a lot of entrepreneurs are looking, and they're probably like, "Okay, I'll build an app, but what's next?" Like any tips for marketing or any lessons that you've seen along the way? First of all, what we talked about with prompting, over communicate. Over communicate, be resourceful. The Replit environment gives you a lot of tools like the logs and things like that. Try to overcommunicate with the agent. I think that's the first tip that I would give. Even if you don't have you don't want to learn prompt engineering, just just be overcommunicative. And the other thing is, like I said, grit. Like just just, you know, not quitting after 6 hours is is very differentiating. Actually, most people just quit. And so, just keep going and not quit. >> Okay, so building something is no longer the bottleneck, and you can go from idea to product in days. This means the only thing that actually matters now is picking the right idea. I talked to the CEO of Opus Clip, and he gave me one of the best frameworks I've ever heard for finding a new niche that actually works. The market is becoming super competitive, and there are like for most of the markets or industries, there are like more than 20, 50 AI tools out there already. Um so, the first principle I would say is to find a really niche business. Oftentimes, many people don't really understand what is a niche. For example, is restaurant business a niche? No. Is Chinese restaurant a niche? No. Is um Cantonese Chinese restaurant business a niche? Still no. Um you have to drill down. Uh there are, you know, so many different types of Cantonese food. Um and even for one specific vertical, um are you going to target to be like the um the Chanel of Cantonese restaurant or the Zara of or Uniqlo of of Cantonese restaurant, right? So, the different pricing will match to different um markets, different ICPs as well. So, pick a really really really small niche that you cannot further segment it to start with. Um that's >> Wow. That's That's a good one. Yeah. Yeah, you have to ruthlessly um segment your niche. The second thing is, ideally pick something that's boring because the non-boring, the cool ones are definitely 10x or even a more 100x more competitive. You probably don't want to work in those areas. So, you need have a passion to build something for a boring industry, but you solve a really big problem. And furthermore, ideally, there are a lot of existing services. I think we are actually, you know, changing from we're redefining the term SaaS. And previously, it was like software as a service. Nowadays, it's becoming service as a software. So, in that specific market niche, boring market niche, are there existing services um done by like agencies, um by freelancers, um or some internal tools, or some hacky solutions, imperfect hacky solutions. Um that is your opportunity to tackle. Um so, go through some boring task, go through some niche task, and go through some service task. Um so, these are the my first principles to find a new business in 2026. >> Young says, "Pick a boring niche." Here's what that looks like in real life. I have this amazing story. So, 2 days ago, my kids are on the school break, and we went to a hotel. And my daughter asked me for coloring sheets with cute deer and cute unicorns. I'm like, "Where?" And so, I went to Google, Googled all of these sheets, and then I saw the website that told me I can generate 30 images like that for you. Choose your child's age, choose the salad they like, pay me 50 cents, and I'll generate 30 pages of ready for you to print. And because we were in a rush, I'm like, "50 cents nothing. I'm just going to pay that." And I paid it immediately. I got my coloring sheets, and I'm like, "That's your real business. That's the real opportunity." Here's the CEO of 11 Labs with a specific one that almost no one is doing yet. It will be voice specific, but I think it's so, so early that I think it's it's a huge one. It's um There's definitely a lot of the infrastructure being built for the voice agents. We We We We build it, but other companies are are are too. Um and I think there is a big gap between voice agents and then actually deploying them in a lot of those businesses. And you don't have to have the engineering expertise to deploy those voice agents. The platform now frequently will support a relatively self-serve manner of taking it, but you can easily take that voice agent and deploy that in a specific domains. And most of the businesses in the world still don't know don't know about it. If it's, you know, not um venture scale business and you just want to make good money, I would try to take those voice agents and go to um let's say local doctor's office and help them appointment schedule for for for the dentist so they can take appointments more easily. And they can then focus more on their work instead of nurse doing that in between or missing appointments. That's actually one of the most common I don't know the percentage, but so frequently those those appointment don't get booked because there's no one on the phone and can take them. Um you can go to local mechanics and help them take appointments. And I think there's all of these require slight variation of the domain piece that you need to know. Um all of those businesses are in thousands to tens of thousands of dollars per month if you get it to the few. Their infrastructure is there, you just need to bring it to to to those domains. >> yeah, it's like B2B automate businesses with AI. >> Yeah, and small businesses all over the world. >> a coder. You don't have to be the coder, you just need to spend the time, call them and ask or or or go to them. Martine just gave you a real business model. Go to local doctors, mechanics, make the offer, get the first clients. But here's the part people always underestimate. You get the clients and then what? Where do you actually keep them? Not on Instagram, not on TikTok. Once the algorithms changes and suddenly your reach drops, your leads disappear, and you're starting from zero again. That's why for me, email has always been the core. It's the one channel you actually own. Nobody can take it away from you. And if you look at most businesses, that's where a big part of their revenue actually comes from. So, before you even scale, before you start adding complexity, I'd set this up first. This part of this video is brought to you by Omnisend, and honestly, it's one of the first tools I'd plug into this kind of system. What I like is that it's not just email. You have everything in one place. Emails, SMS, push notifications, automations, so you're not stitching together five different tools and hoping they talk to each other. You build one system once, and then it keeps working in the background while you focus on getting more clients. And the numbers are pretty wild. Their customers see around $79 back for every dollar spent. That's one of the highest ROIs you'll see in this space. Also, if you're already using something else and it's getting expensive as you grow, they'll actually migrate everything for you for free. Real people handle your contacts, automations, flows, SMS numbers. Usually takes three to five days, and your old system keeps running the whole time, so there's no risk of losing your clients. They offer 24/7 support by real humans, and what's actually valuable is that their average response time is under 5 minutes. If you want to try it, you can use my code Marina. It gives you 30% off your first 3 months. I'll leave the link in the description below, and now let's get back to the video. Okay, you have the niche, you have the tools. Now, what does the actual 90-day launch plan look like? Alex built his field to 200 million in annual revenue in 9 months. He raised money before he had a single dollar in revenue. But when I asked him how to start from zero today, he said he'd do it completely differently. Focus on bringing like the first dollar by day 30 of product developments, and maybe 1 million ARR by day 90, and then decide if someone needs VC funding or doesn't. I mean, a lot of businesses can really scale to tens of millions of dollars today profitably with AI. And for such businesses, there is no need to attract VC funding. I can give you a very simple example. So, there are so many uh websites which just allow to make professional photo shoot, like for basically for for passport. Many of them make tens of millions of dollars. None of them are going to be Yeah, none of them are going to be venture capital backed business. Well, and you said something $1 million by day 90. AR, meaning like 80K a month. 80K a month in 3 months, that's a lot. And so, the the playbook to achieve that is basically generate ads and launch them and test whether whether they're landing with your target audience. Paid ads are very difficult today. I think a lot of distribution come through organic social media and creator integrations. And just make sure that by day 30, uh there is monetization in place, and then there is a way to constantly grow to a revenue to, let's say, one a million AR by day 90. Many successful uh companies scale very quickly today. Any tips on landing first customers in the first 30 days? Initially, at at least last year, Twitter has been the social media where the distribution starts from. It starts from like small communities, then it goes to AI news pages on X, then from AI news pages on X, it goes to Instagram news pages, then from Instagram news pages to creators, then it goes to Telegram and like other social media. That's what we have seen with Higgsfield and with many other products as well, that they went through the same sort of journey of popularity and news through various social media. But, it all originated on X. I think now it's being kind of changed. Today, a lot of hype like a lot of companies, they sort of try to use X to boost their product. A sort of signal to noise ratio just drops. Twitter becomes less relevant, but still it's it's the main place for new AI products launch. Alex just said you don't need investors. And here's what's wild. Bill Gurley spent decades backing the biggest companies in Silicon Valley. And here's what he told me about why this is actually the right moment to move fast. If you're a high agency individual who's who has confidence about driving their own career path, these tools are actually a jetpack. Like you if you if you want to proactively learn, there's been no time in the history of the world where you can learn faster. And you know, it's it's AI, but it's also podcasts like the stuff you do. There's interviews all over YouTube. It like you can learn faster than ever before. The thing to do right now is to run as fast as you can. >> If you have a friend who keeps saying they want to start something but don't know where to start, share this video with them. Okay, one last thing. Because after all these conversations, I kept thinking if the opportunity is this obvious, what's the real difference between people who make it work and people who give up? >> There are a few few things I could say here. One is launch, launch, launch. Just keep launching. Go launch on Reddit. >> iterate. >> Even the same product. Uh like make make another >> Another tweak. Another tweak. Show it in a different way. Iterate on your messaging. Do another video. Uh try to reach out to influencers to partner with them. Go on a go >> Ruffler, I launched it three or four different times with different messaging and different things. >> Interesting. So, the first three times didn't work? Uh first few times didn't work. Uh I think when we got on Hacker News the first time uh it was when I said uh you can try all these different languages and I listed the languages. Try Python, uh Ruby. That was before AI. Uh and that was a hit because of the because of the title change. And so again, it's grit, relentless resourcefulness, and just iteration iteration iteration. After all these conversations, this is what really stayed with me. The barrier to starting a business is basically gone. You don't need a team anymore. You don't need to know how to code. And in many cases, you don't even need funding. But what you need is speed. Because the people who are winning right now are not necessarily the smartest or the most technical. They're just the ones moving faster than everyone else. And that's really the shift. So, if you zoom out, the playbook is actually simple. Pick a boring niche, build an MVP in 24 hours, get your first dollar by day 30, and use organic content as your distribution. If you want the step-by-step checklist for everything you just heard, it's in this week's newsletter. Link is in the description. And if this gave you the plan, the next episode gives you the speed. Alex Mashrabov breaks down exactly how to go from zero to 1 million ARR, annual run rate, in 90 days. Oh, watch it right here.