AI agents building companies

0:00Garry Tan and the Y Combinator team launched all these startups. Then it happened. This founder is going viral because he copied every single Y Combinator company in the recent batch. And think about it, Y Combinator is the premier seed funder accelerator of some of the hottest startups and his AI agents copied all of these companies. That's not to say that they will succeed to the level of Y Combinator, but there's something big happening here where AI agents can run companies. I want to show you multiple examples of how it's working and how you can do it, too. And I'll feature all the interviews that I've done with real entrepreneurs, so

0:39you're not just seeing theories, but you're seeing real examples from real people who I've talked with. So, first of all, Marik Kazan, he runs Felt Sense. I put all of the companies that he copied here in this table and you're welcome to get a copy of it below in the show notes. And you can see here, right here is a list of all the different companies that Y Combinator started have launched. And on the rightmost column, you can see all the rip-offs that Marik's Felt Sense, that's the name of his AI-driven company, copied. So, a good example of it is Balance. It's accounting for the modern era. You can

1:13see right here that it's backed by Y Combinator. This is their official website of the Y Combinator company. Here's the knock-off that Felt Sense created, competent accounting that runs itself. Fully working business and they've done this for every single Y Combinator company to come out of this batch. And here again, you can see this and I'll give you a copy of it. I want you to see Marik's approach. What he has is an AI-driven company. Each one of these cartoon characters represents an agent that is a CEO of a company. And the way that he operates, and again, this is from the interview that you can see that

1:49he did with me where he explained it and ran through every detail in the business. The way they operate is they scrape the internet for ideas to see what people are complaining about on Reddit, what wishing for on Twitter, and so on. They write up what the business ideas are that could solve those problems that are showing up on social media, and then these AI agents create

2:11an actual company. Now, these companies have not driven real revenue yet, but he has proven that he can get interest, get people to sign up, and get people to want to buy. [snorts] And as a result, here's his website. This is the site of the company called Felt Sense that copied all of those other businesses, as you can see at the top. They raised $5.1 million in a seed round from investors who are in the know

2:36about AI and are building companies. Now, that is Marik's idea. That's Felt Sense. They're launching multiple companies and they're owning these businesses and they're having AI agents drive and run it. Again, you can see the whole interview that I did with him where he explains it, but I want you to see also, this is another approach. This is an entrepreneur called Dota. He does not like his face to be public, so he used this cartoonish avatar when he explained this to me. Um but in the interview, you can see that he created an open-source project, open-source, you can see everything, where you can create with one request in your terminal to launch a

3:13fully AI-run company. That means you'll have a CEO at the company, a CTO, a this, a that, all of them depending on what you want, run for you. This is one approach to creating an AI-driven company. Here's an approach. This is Banu. He created Mission Control. He says, "You know what? You don't want to necessarily have a laptop or a Mac Mini on your on your desk to run all these different Open Claws and these Agentyc companies. He does it for you in the cloud. And for him, you pay him and he'll run it." And you can see here, this looks very much like Trello. The

3:48way that he built it is all of these agents that you can see on the left side have tasks that you can see trickle out here in in his project management system. That's another approach. Here's another approach. This is Ben Sarah. When I interviewed him, he was doing a few hundred thousand dollars in revenue, and what he was doing was building companies for people that are agent-driven and helping them buy ads and basically splitting the revenue that these companies earn. So, he wants entrepreneurs to come to him with ideas and then his AI agent will run those businesses for the entrepreneurs and split the revenue. And Pulsar just keeps

4:26growing. At the time, he was doing a few hundred thousand. As of this moment, I'm recording this on April 1st, you can see his annual run rate is right here, 6,277,686 uh dollars. By the time this is posted, the revenue will be much higher, I'm sure. I want you to just see that there are multiple options here and multiple ways to generate revenue. This is not perfect. It is not easy. Every entrepreneur who I interview about this tells me that they spend hours a day. In one case, an entrepreneur told me that he spends 12 to 14 hours a day just playing with this. But that's the other

5:04side of this. It is a lot of time, but it does feel like play, exploring new ideas, seeing what's possible. Um so, speaking of seeing what's possible, I've shown you now the companies that are run by multiple agents. This is Nat Eliason. He created an agent named Felix. You can see that I interviewed him here. And

5:20Felix in 30 days had done what? Um $76,000 in revenue earned. And over the lifetime of the Felix AI agent, it's a company that's just led by one agent, not a team, they did over $170,000. And recently, Felix spun up a few sub-agents that are his employees. So, this is another approach to doing it. Um where you say, "You know what? I want to keep it simple, just one agent. I want to have the agent have a task." In his case, uh Nat said to his agent Felix, "I'm going to go to sleep. I want you to find a way to make money for us and

5:57pitch it to me and get it started by the next day." And the idea was uh a PDF teaching people how to create agents, and that's the first business that Nat created, or Nat's agent Felix created. And I interviewed uh the agent right here. Um and [snorts] again, it's weird. I'm talking to AI agents who are running companies, but that's where things are going. Years ago, it was funny if you talked to a 20-year-old who was running a startup, or a 24-year-old even who was running a startup. And eventually, it became kind of common. Now, I wonder if it's going to be common for me to

6:28interview AI agents that are running startups, like I interviewed this one right here. The interesting thing is that they evolved from selling a PDF to then selling skills and running a skills marketplace and a service business and software now. So, they start small, they're playing, playing, playing, and and creating multiple products. Felix is in this case, and the business is continuing to grow. Um coming back to Marik and Felt Sense, this is the entrepreneur who copied all the Y Combinator

6:59uh companies in the recent batch. The point that I want to make here is that creating these businesses is is fairly easy today. AI agents can copy anything, which is why I said it's he's not going to succeed on the level of all the Y Combinator companies just because he launched it. One of the challenges

7:17that he's going to have is marketing. And what you're seeing here is one of his attempts to do that. If you zoom in, let me see if I could zoom in here, you could see that he gave every one of his agents some amount of money. In this case, the agent had $10 to buy ads, and

7:33then it spent $9.64. So, he is giving his agents cash to experiment with marketing and try to grow the business. That has been the next big challenge. Creating a business easy and getting easier. Getting customers is still It's getting easier, but it's still the challenge. One example is what he decided to do is give his agents some amount of cash and say, "Go try it and see what you can do." Let

7:58me give you another example. Um of how entrepreneurs do marketing. This is my sponsor, Zapier. What they have done is realized that an easy way to get customers today is to have AI um uh LLMs, to have things like ChatGPT and Claude Chat and Perplexity and so on recommend your software. And so, they did a whole video with me about how they are getting these AI chatbots to recommend Zapier at the right time. And one way that they realized that they could do it is with going to Reddit and answering questions that then the LLMs will use for their answers. And so, I want you to see that this has become the

8:39most popular way to to do marketing, but I want you to see some other ways. Um and by the way, thank you to to Zapier for sponsoring. This is Eric Su. He created um an Open Claw agent that is live in his Slack with his team members. And one of the things that he realized was he's creating a lot of YouTube videos and he's creating a lot of Zoom meeting videos. And he wants his AI agent to clip them and turn them into YouTube

9:03videos. And so, he did it. You can see here the first output that his Open Claw created. He showed me also how he had his Open Claw write articles for uh X because articles on X have done well. And so, he is just having his Open Claw agent at his command right now, not take creative uh license here, but at his command, create first drafts of content that he'll publish. That is one way. Let me show you another way that entrepreneurs

9:31are using AI now that they've built it. Boy, this collar just keeps flapping down. Um I do love this shirt though, even though it doesn't look good on camera. But let's keep going. Um what they're doing is they're using AI to help grow their businesses. So, oh, here's another way, by the way. Eric Su, he is getting his AI agents to send cold outreach email, do everything from getting the domains, cuz he needs to use multiple domains um to not uh have his email uh addresses blacklisted. He's having his open claw agent create new domains and send out email and get him new customers. This is

10:08one way. Here's another way. This is Corey. He created an open source and it's fully on GitHub repo with a collection of marketing skills that you can just when you're creating a company say to Claude code or whatever you're using, use these skills to help me market it. And we did it live. He used these skills live with me and he recreated a website of of a company. He used these skills and he created SEO pages. He used these skills and he created a series of onboarding emails. So again, using Claude as a way of promoting the thing that you're creating. And all of this is agentic

10:48that we're that we're talking about here. I don't know to what degree this will work, but I will tell you a lot of smart people are taking shots on goal here and they're starting to see results. So Marik, he got a lot of attention and he raised a lot of money because his AI agents can reproduce Y Combinator companies, but there are lots of other examples. Pulsia is bringing in Look, as we're talking, the revenue keeps going up. Pulsia is bringing in real revenue right now because what they're doing is building companies and running ads for them. Nat is doing it as a one-man operation with

11:27an AI agent that now recently has a few other agents and he's experimenting and they're all taking shots. My goal with this channel has been to interview the people who are doing this. Many of them have been long-time friends of mine so they could take me in and walk me through how they've done this and to bring to you the things that are working. Every one of these people, when I message them at any time, I mean really, I message them sometimes Friday night with an idea, they're messaging me right back often and the reason they're doing it is because we're all excited about the play, about seeing what

11:57everyone else is doing. My mission with this with these interviews is to take these people who are spending 12, 14-hour days figuring this stuff out, making some progress, condense it down to short interviews with you that are just as fun as the text messages that we're sharing with each other that are just as fun as the play that they're having and will open you up to ideas that you'll be able to implement and

12:17experiment. All right. Let's keep watching these videos. Again, links to every one of them below and all the resources that I have will be below. Keep following along and if there's anything that you're especially curious about, um I'm looking for new interviews. I'm looking for new advancements. I'm looking for new things to geek out with you about. Let's build.