AI CEO: How to build a $1B Company in 2 days | Amjad Masad @replit — Silicon Valley Girl Podcast

Amjad Masad: Yeah, something like that. It will cost you something like that. But I would spend two, three days on it. I think you'll be able to get it.

Marina Mogilko: How can I learn to be better at prompting?

Amjad Masad: We have a YouTube channel. We have a great developer relations person who creates a lot of content. His name is Matt. And so we try to train people on prompting and the underlying systems. So Replit has a DNA of education. So when we were talking about a billion developers, these billion developers need to learn. It's not going to come for free. So there's a learning curve associated with it. And you need to be resourceful. So you need to go to YouTube, search like "how to prompt." You need to spend a lot of time and practice by building, changing your style. Some people go to OpenAI for example and pick O1 or right now it's GPT-5 with thinking, and give it the idea and tell it, "Hey, I want you to structure it into a really great prompt for me," something like that.

Marina Mogilko: Yeah. So I think Paul Graham, the founder of Y Combinator—you know, the best accelerator in the world—wrote this essay about being resourceful. He talks about the qualities of founders, and one quality is being relentlessly resourceful. So they're able to find resources to unblock themselves and not hit a wall. Because I think a lot of what entrepreneurship is is finding all these walls and really driving through them.

Amjad Masad: And the way to do that is—think of it like a video game. In a video game, especially open world video games, you're often running into these problems where you don't know how to get to the next level. A lot is creative thinking, kind of moving around and finding the clues. And I think entrepreneurship and building software is similar to that.

Marina Mogilko: Can you tell me what changed when you decided to start your company? What was this thought that was like, "I need to build this"?

Amjad Masad: So there's starting the project versus starting the company. Starting the project was so obvious to me—programming is hard. We need to make programming easier. That's like a very technical sort of view on things. Starting the business was less obvious because I had worked at startups. I knew how painful they are. Like, I worked at Codecademy for example when I first came to the states. It was based on the open source version of Replit. But I saw how difficult it was and it was really painful—a lot of hard work. So Replit was still a side project and started growing. And we really—I didn't want to start it into a business because I actually tried to sell it to Facebook where I was working back then. I wanted to stay there. I was very comfortable and happy there. It's a scary feeling to leave your job and go heads down.

But it was kind of de-risked because we had a platform that people really loved. We had 100,000 users plus a month. So after a lot of deliberation and really trying to think what matters to you, what creates meaning in your life, and serving our customers and really achieving our mission—helping people create businesses and all of that—felt very important. So I think it's about meaning partly.

Marina Mogilko: It feels like with this tool there are two problems in entrepreneurship used to be three—like coming up with idea, building product, and marketing. Now you're left with coming up with ideas and marketing, right?

Amjad Masad: What do you think is going to happen when everyone's building an app?

Marina Mogilko: Yeah. So when everyone's building an app—when that is—it's still a skill, right? It's still like we talked about. It's still a skill that you can develop. And I think grit is very important. So resourcefulness, grit, like not quitting, like not quitting after six hours, like spending another day or two on it at least. I think domain knowledge is very important. So if you have excellent domain knowledge on YouTube, you need to imbue that. You need to give that domain knowledge into the agent. You need to prompt in a certain way so that you're downloading your domain knowledge, and that is your competitive advantage. But at the same time, what OpenAI models are training on, they're so much better at defining what a good YouTube video is.

Marina Mogilko: I think you still have tacit knowledge that is not necessarily expressed in all your videos and all the content out there. That CFO at the VC firm has a lot of knowledge and skills he built up over the years that he can make into an app that you can't find on blogs and you can't find online. And so I think every one of us as we go through life, we build up a lot of experiences that LLMs do not get to experience because they're not embodied. But do you think there will ever be a time when AI sees the problem, comes up with a solution, codes the app, and doesn't need a human?

Amjad Masad: You know, I might be a bit different in the Silicon Valley context in that I am quite skeptical about the AGI vision. I think we can build extremely competent agents, but you would always need the human as a driver because I think that the way large language models work is they train on the entire corpus of texts on the internet—text and books, all of that stuff that is text of the past, of what has happened, right? What people have put in. But can it come up with novel ideas, creative ideas, ideas based on what's changing in the world right now? Because they're not always learning. They're not continuously learning. They're within this closed box that is their training corpus, right? It's almost like a library in a way. It is a library. A library has a lot of ideas and it can remix and mix and match ideas. But a net new idea is something that I think still humans have a special place.

Marina Mogilko: I love your answer. But so the fear inside me wants to push back by saying that, you know, all the ideas already exist in the world. Like when we're talking about new movies or new books, they take a pre-existing idea but change characters, change circumstances.

Amjad Masad: There's always a novel idea. So think about Bitcoin. Bitcoin was based on a history of 20 years of people trying to build digital cash. And in the references it's referencing hash, proof of work was existing. Proof of work—the idea that the machine is solving these cryptographic problems in order to secure the Bitcoin network—was actually originally invented to fight spam. So there was the spam problem of emails. We didn't have spam filters or AI. And so when I'm sending you an email, I solve a cryptographic problem that is expensive to show you that I'm not a spam agent, right? So Satoshi Nakamoto took all these ideas—and you're right, existing ideas—and put them in a new package. But he added a novel idea, which is how to solve the double spend problem. And this is the blockchain, right? So I think it's easy to think that there's nothing new under the sun. I think that's the expression. But I think if you look carefully—if you look at what Einstein did, right, with his theories—there's always one novel insight, one really strong novel insight. It's almost like there's divine intervention. There's something spiritual about having a really novel idea. And I'm sure you've experienced it in the past. I think that is fundamentally human. I don't know where it comes from.

Marina Mogilko: Yeah. And also this feeling that you're the one to bring it to the world. I feel like AI lacks that. Okay, let's see. Let's see what's going on here. Should I go to overview and try again? No.

Amjad Masad: So I hit redeploy. It's asking you to redeploy. So it'll take a second to deploy. I'd be curious. So fixed session decentralization problem. Okay, while it's deploying—you said the gap between a Replit user and a senior Google engineer will disappear in two years. Should people still learn how to code, or what's going to happen to the engineering job?

Marina Mogilko: I think the engineering job will continue to exist, especially in very domain-specific areas. That's basically what we're talking about. There are a lot of things that are not very well represented in the data. If you're a platform engineer at Google dealing with a billion users, there are knowledge and things that you understand and have learned on the job that LLMs do not know because no one's written them anywhere. It's this tacit knowledge.

And so I think those engineers will continue to exist. If you're an engineer at NASA and you're building fault tolerant systems, if you're building provable systems—I don't want my Tesla autopilot to be vibe coded, right? Like there are a lot of life and death systems that we want engineers that are very low-level and very almost mathematical about it to exist. So there's a lot of situations in which engineers will continue to exist.

But if you're a product builder, I would say just go ahead and build the product. Like don't wait. If you need to learn coding along the way, learn it. But your mission is to build the thing. So I would start by building and, like we said, being resourceful along the way goes a long way.

Amjad Masad: So what should engineers who are building apps like this do now?

Marina Mogilko: Um, let's see. Where is it? There we go. Oh, there we go. Okay, sign in. Okay. We had a new issue, and it's like Google authentication, but that's how software engineers work. Okay, this is this iteration. Got it. So you think we'll solve a problem, and then get a new problem? But do you think it will ever get to the stage where I don't have to do this?

Amjad Masad: Yes. Because it could actually run everything and test everything. Yes, so we're working on that. Is there something you can talk about it, or not yet?

Marina Mogilko: Usually not yet. But let me give you some hints. Okay. Every vibe coding platform today automates generation of code and all of that stuff, which is great, but leaves a job for you that is actually very routine and uncreative and annoying, which is quality assurance and testing.

Amjad Masad: Exactly. QA.

Marina Mogilko: So we're solving that.

Amjad Masad: Okay. All right. You're launching something right? Is there a lot of pressure on the team? How are they handling?

Marina Mogilko: Yeah. So we have an offsite next week where we're going to LA on the beach. At least that's like a sort of relaxing environment and we're going to be going really deep and working really hard. And then the week after that we're going to be coming here. It's called sprint week and we do it before every launch. And people typically work 14-hour days non-stop.

Amjad Masad: You too?

Marina Mogilko: Yeah.

Amjad Masad: Your wife too?

Marina Mogilko: Yes. Yes.

Amjad Masad: Who takes care of the kids?

Marina Mogilko: Okay. There's a question coming from all the entrepreneurs who have kids: how do you balance this? So we talk about work-life balance. We talk about work-life harmony or work-life integration. And so, for example, we'll have the nanny bring the kids to the office. So they can vibe code. They can vibe code. I sometimes sit down with my kid and do a little bit of coding. But so we can see them, right? And next week for the trip, everyone here could bring their families as well. So we try to create that integration.

Amjad Masad: Do you think it works honestly, or do you feel like you're missing out on your kids?

Marina Mogilko: I don't feel like I'm missing out. Like, honestly, I don't feel like I'm missing out. So just having a bit of freedom in your schedule. I spend like mornings with them. I wake up at 7, 8, and I get to the office by 10:00, right? But I stay late here. We have dinners here. But I have these two hours in the morning at least. Sometimes I'm there for dinner as well. And on the weekend, I'll work a little bit—like Sunday afternoon. I'm always on my phone working obviously and responding and taking phone calls and things like that, but I feel like I'm present.

Amjad Masad: What about your wife? Similar schedule?

Marina Mogilko: I think, yeah, no, she has a less intense schedule. I think the CEO schedule is a little more intense. And I think this is probably something you can relate to. I think being a mom there's like a more sense of guilt that I don't really feel all the time because they're back from school right now, right? But I also would love to do this. Yeah. But I also feel like all things considered, we feel fairly involved in their lives.

Amjad Masad: That's good. Talk to me about your mindset. Like you're coming up with all these new things, but at the same time we have Lovable, Cursor. We have all of these tools. We're going to let you do the same thing. How do you see yourself being different? And how do you survive mentally as an entrepreneur?

Marina Mogilko: So we're going to be the first to do what I just talked about, and it's launching in September. Yep. And last year we're the first agent on the market. And so we're always three to six months ahead. I think this feature that we're building is probably going to be a year ahead of anyone else. And that's because it's built on 10 years of innovation, infrastructure innovation. Like every app that you're building, every workspace is backed by a cloud virtual machine built on a file system that we innovated. We even patched the Linux kernel to make things work for Replit. And so all that infrastructure allows us to always be ahead.

We'll come up with an idea. Sometimes it takes two three months to build. If you want to build the idea from scratch, it'll take you two three years to build. And so I think Replit will actually start to diverge pretty soon. All these applications kind of look the same because they all generate a website. But when you talk to developers that are using Replit, often people moved from the other platforms, and they're saying that, yeah, I mean these tools got me a pretty website pretty quickly, but you know, a month in I'm just blocked. I can't manage my database. I want a place to do storage for my files. Replit ships with a database, has an object storage component, has authentication component. Like I know you're trying to set up Google authentication, but you can also ask the agent to implement Replit authentication. We have a built-in authentication system. We built up this massive amount of infrastructure, and within the next 6 months to 12 months I think it'll be really obvious how Replit differentiates.

Marina Mogilko: Have you seen any big mistakes that people make when building something with Replit? Because 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—overcommunicate. Overcommunicate. 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 want to learn prompt engineering, just be overcommunicative.

Amjad Masad: And then on marketing, I think that's the next big bottleneck for entrepreneurs, right? Let's say building a product becomes easier and easier and easier. How do you go get it to market and communicate your value proposition? All of that.

Marina Mogilko: Exactly. And like, why would someone buy something like this if they can just go to Replit and vibe code the same thing, right?

Amjad Masad: Yeah. There's definitely marketing is a big part of the answer. But also, like I said, domain knowledge that you have. Like just think about the things that you know, deep in your heart of hearts, that not many people in the world know. And the other thing is, like I said, grit. Like, just not quitting after six hours is very differentiating, actually. Most people just quit. And so just keep going and not quit. Like, I've been building this business for 10 years. Before that, back in Jordan, I had this idea when I was 22 and I started working on it. Actually, there was an open source project called Replit back in 2010, and I just didn't quit. I knew this was kind of big. I still know it's going to be a trillion dollar company at some point in the future. Right now we're like a