So, the vision became anyone with an idea should come and be able to build software. He's a teacher, he's a domain expert, but he doesn't know how to code. But using Replit and AI, he was able to get an MVP, get a few first sales, and now the company is worth half a million dollars.
I take pride that we're very efficient. We have a $9 billion valuation with 200 people. It hasn't happened in history. Like we have a 25-year-old that's retired.
Then the competition came.
How do you react to that?
This is Amjad Masad, born in Jordan. He was rejected by YC four different times. His company, Replit, went from a struggling startup to one of the fastest-growing AI companies in the world, reaching a $9 billion valuation, officially making Amjad a billionaire. And it's one of the leaders in the web coding space. Just a few years ago, Replit was generating only $2.5 million. Then it jumped to 250 million, and now the company is aiming for $1 billion. Web coding is the application of AI to software development. On Replit, anyone can build apps, websites, and software products without writing a single line of code. Amjad's right-hand man and Replit president is Italian. I've already interviewed him, and in that conversation, we focused mainly on AI and the technology behind Replit. In this interview, we're going deeper into Amjad's story and exploring how AI is reshaping the economy, entrepreneurship, and the future of work, creating entirely new phenomena like the rise of solopreneurs.
I think the first stroke of luck that I had was that my father bought a computer when we were very young. That was 1993. I was 6 years old back where I'm from in Amman, Jordan, and no one I knew at the time had a computer. One of my earliest memories watching him unbox a computer, pulling up a huge manual, and like reading it and trying to type things. I sort of immediately got it. Something about the computer fit the way I think, and I got good at it pretty quickly. It took a minute and a half for my younger brother to learn math, and it went so well that now he works for me here. And then I built like my first business when I was maybe 13 or 14 years old. I don't know if in Italy you had this, but like around late '90s we had a lot of internet cafes and also LAN gaming cafes. We used to go play Counter-Strike, you know, things like that. I used to be addicted to these games. It was a place full of computers, but they were using all the staff to manage it and like write down everything down paper. So, I sold them on this idea I could build you a piece of software with a client server management accounting software and create gift cards. You could do all this marketing stuff. They said, 'That sounds great. We'll buy it.' And then I went away for two years. It took me two years to develop it because I was just like learning how to do the stuff. I made hundreds of dollars which was infinity money at the time for me. I took my entire class to McDonald's when it opened up in Amman. I was balling.
That was a lot of fun just making your first money and you know this stuff. Talking to a lot of entrepreneurs, it's just like an intoxicating feeling and you're always chasing that high again. In college, I got interested in programming languages. You know, I was taking Python and Java and different programming languages and I thought it was such a pain to have to install a programming environment. So, I was like, 'Why isn't there like a website where you can just like say I want to program in Python and just start coding?' And that didn't exist and that also took me about two years to crack. And I built the first open source programming environment. That's kind of the basis for what becomes Replit 10 years later. And a bunch of companies in the United States were starting to online education. So, that was a big thing in like 2011, 2012. One of them is Codecademy and so they used my software to teach programming in the browser. So, they called me. They sent me an email and they're like, 'We're using your software. Like we'd like for you to work with us.' And I started working with them remotely. They paid me $15 an hour which I thought was amazing.
And then they're like, 'Hey, we want you to join us.' I'm like, 'No, I'm not going to join. I want to start my own company.' I already had the idea for Replit back then. But they flew out to Jordan from Silicon Valley to meet me there and to convince me to join them. Every day they'll give me an offer. I got to say that's not enough. I want to be a co-founder. They stayed there for a couple days. I was driving them to the airport and then the CEO gave me an offer that was like what I expected, and I signed over there. They got me an O-1 visa. I was the first engineer. They called it founding engineer, and I came to New York. I was getting paid nothing because I negotiated a very high equity, but not good salary. You know, I've seen my family and it's like salary is just it's not how you build wealth. So, they were paying me $70,000 in New York City. That is hugely painful. But you know, when you're in your 20s, it's just like doesn't matter. You can sleep on the floor. You can eat whatever. Oh, when I got there, I literally couldn't get an apartment, but I still didn't care because I was going to spend all my time at the office anyways, and you know, they were generous enough to give me this like huge equity. And that paid off cuz the company sold for half a billion dollars like 10 years later.
But even back then during Codecademy, I started seeing what the impact of giving the power of coding to more people could do cuz I started seeing entrepreneurs who had very little coding background, like a fitness trainer coming to Codecademy at the time, learning a little bit of coding, building an app, making millions of dollars in the App Store. Like, wow, this is really powerful. This is transformative for people's lives. So, the idea for Replit starts forming, which is a place where very easy place to code, but also very easy place to deploy and host, and help people go all the way from an idea to an app. And so, after Codecademy, I went to work for Facebook a little bit because I was really interested in a few things. One, React JS had just come out, and React now is the top, you know, UI library language in the world, and the inventor actually is the head of product at Replit now.
Yeah, he's sitting over there.
So, in 2016, I felt like I had enough stability, saved up a little bit of money, and my wife is a designer, so we took that open source project, we dusted it off, called it Replit. It was like an IT domain, an Italian domain, which is really hard to register. You had to know someone in Italy to register it. We still own that. So, all the email addresses, if you realize,
Yeah, it was really cool to do these domain hacks back then. We get the site online and then we get users immediately because the need was there, but still there was a lot of road map for us to build, right? So, the vision became anyone with an idea should come and be able to build software, right? Very ambitious. We went to investors and they're like, 'This is crazy idea. It's way too ambitious.' The thing you hear about Silicon Valley from the outside is that you can come in with a crazy idea and be able. Typically, you still need to be a PhD or to be a Stanford dropout. So, if a kid from Jordan, like coming in with a crazy ambitious idea with no track record, it's really hard to raise any money. I'm the kind of person who doesn't quit, so we kept going. We started selling to schools again cuz that was the easiest place to make money. Fast forward 2 years to 2018, we kept building the platform and I was blogging about it. People call it building in public now.
It wasn't a thing back then, but I was always interested in kind of talking about what I'm building. Paul Graham, the founder of Y Combinator, the premier startup accelerator in Silicon Valley, he reads my blog because it got onto Hacker News and he's on Hacker News all the time. They reach out to us and they tell us, 'Hey, you should come and do YC.' And that was the first big break where we got like a stamp from a like a real Silicon Valley investor.
We went to YC in 2018. Sam Altman was running it at the time, so I got to know him, got to know Paul Graham very well. YC had rejected us four times before that.
And so, we got into the back door, but having had that experience actually gave us a better exposure. The story became a lot more interesting. This company that was rejected four times, it got into YC. And so, during YC and the few years after that, we built out the platform. So, now you have the development environment, you have the deployment environment. And I always felt like AI was going to have an impact on coding. In 2020 and 2021, OpenAI had come out with GPT-2 and then GPT-3. And this is the thing about being in Silicon Valley or being plugged into the kind of what's the latest and what's happening in technology. I saw it very early on and said, 'Okay, this is what we have to bet our company on.' It was very primitive though. So, the first few products that we built were not that useful. It would generate some buggy code and people would like play with it. It was like, 'Oh, that's kind of cool toy.'
Can you tell me a little bit about the numbers before the vibe coding trend?
We always had a lot of users. So, we had millions of users.
Because it was a viral product, right?
It was a viral product and that helped us raise money.
So, investors want to see a graph go up.
At the YC Combinator demo day yesterday and all the charts were like this even if it didn't make sense.
Yeah, like they'll come up with a whatever random metric like hours or whatever, you know? But the revenue was hard because we weren't getting people far enough in the process. Coding was still too hard for most people to stick with it to get to a point where they can deploy an application that's actually useful or making money. And I always had this mindset of like it'll take off at some point. It'll take off at some point. And that's what I was able to sell investors is that I will get better and the moment it does and it's actually easier to code, we're going to go like this. And that exactly happened. We went from 2.5 to 250 million by the end of 2025.
So, that is a really pivotal moment. So, what were the things that made it possible?
So, I met Michele, now our president and head of AI, as part of a deal we were making with Google. So, I was going around Silicon Valley. There wasn't anyone paying attention to AI coding. And I was talking to OpenAI, I was talking to Anthropic, I was talking to Google. I was telling them, 'Look, we have this platform, millions of users, we have a lot of data on coding and we want to partner with one of you to develop like coding models.' Everyone was very careful. OpenAI was like, 'Oh, we're going to work with Microsoft.' Google was like, 'Ah, that's we don't care about that.' Except for Michele and his team. They had like a small research group that was focused on AI for coding. So, finally got to him and his group. We got excited about potentially doing a deal together and we got close, but then Google actually killed it. And the reason is because Google Search is such a profitable product that they were worried about the reputation, and they were worried about Google Search being disrupted.
They denied it, but it's true.
Yes, Michele ended up leaving, and so I convinced him to join part-time. It's like, 'Look, I tried with all these different companies. We have to train our own model.' Which is a radical thing to do in like 2023. But we ended up training a state-of-the-art coding model. And that's when our revenue went from 1 million to 2.5 million. So it was already, you see, it's some kind of a ramp-up. But again, it wasn't enough. And so in 2024, you know, Michele and I were always talking about agents. When are agents coming? Like, agents are the holy grail. And the thing that differentiates agents from chatbots is chatbots can give you a bunch of tasks. Agents can do actions on your behalf. So 2024, we reached an impasse. Revenue wasn't growing. The product was in this middle place where it had a lot of users, but not enough revenue. There was a lot of burn. We had to do a layoff, unfortunately. We laid off like, you know, 30% of the team. Bunch of people left. But we decided in summer 2024 to build a coding agent. At the time there was none. So it was like a visionary moment for us. And so we spent the entire summer here in this office. We had just moved. Just locked in. We all kind of felt like we had a secret that no one else in the world knew because we knew that the technology was moving fast enough. Even the big labs didn't know that the models were capable of that. So September 2024, we launched the first-ever live coding product. It wasn't even called live coding. So that was Agent 1. Now we're in Agent version 4.
So to recap, Amjad fell in love with computers as a kid in Jordan and built his first software business as a teenager. Later, he created an open-source coding tool that caught the attention of Codecademy and brought him to the United States. After working at Codecademy and Facebook, he teamed up with his wife Haya to launch Replit, a company built around a simple but ambitious idea: make software creation accessible to everyone. The company struggled for years. Replit was rejected by Y Combinator four times, and despite having millions of users, generated only about 2.5 million dollars in annual recurring revenue. But Amjad believed AI would eventually transform programming. When he met Italian AI researcher Michele Catasta, the two doubled down on that vision, betting everything on AI-powered coding agents. In September 2024, they launched one of the first web coding products on the market, and within a year, Replit's revenue exploded from 2.5 million dollars to 250 million dollars, turning a decade-long bet into one of the biggest success stories of the AI era.
Then, the competition came.
How do you react to that?
The most important thing is to listen to your customers, right? What do your customers want? What's missing? And the thing is, despite a lot of competition coming in, a lot of the products that we're getting built were not really helping users reach the milestone that we want them to reach, which is build a production app that can scale, that can make revenue. So, all our efforts went into how do you make the agent actually produce good code, sustainable code, you know, things like that. I think where the competition matters is for a long time we didn't do any marketing. The marketing was my Twitter account, my YouTube interviews. And now for the first time, like so much money getting poured in, everyone's betting against our keyword, go Google Replit, there's like a full page of ads, and we're like, okay, we got to take marketing really seriously, and we got to take go to market seriously, we got to take sales seriously. Marketing lagged behind a little bit, but now we have a marketing leader, and we're starting to build up the marketing org, and we're an organization that not only do we have to sprint and get product out really quickly, but we have to get the product into people's hands very quickly as well. So, we have a, you know, 50-person sales team now, they're selling to enterprise. We have amazing customers like American Express and PayPal, but everyone else, the idea that anyone can code, it'll take a while for people to wake up to it.
Now, it's not very common outside the US.
Outside of US is not very common, and maybe hopefully this interview will make a dent in that. But, the main message I would want to encourage anyone watching this, you can code now.
You can build things. You know, I was just at a conference in DC, an entrepreneur walked up to me, and he's like, 'You changed my life.' He said, 'I started using Replit, and it transformed everything about our business. We became a lot more efficient. We started selling more. We automated so many things about our business, and I hired eight more people, and now we're making more money, and I'm doing less in the business.' But, then there are people who start businesses that are very valuable, but also very impactful in the world. There's a teacher who had an idea to build AI tools for teachers. Grade, create assignments, you know, just automate parts of their job. He's a teacher. He's a domain expert, but he doesn't know how to code. But, using Replit and AI, he was able to get an MVP, get a few first sales, raise money, and now a company is worth half a billion dollars.
It's called Magic School.
Since the beginning until now, the real moat for founders was how much they were good at coding and creating products. Now, it appears is different. The more you are an orchestrator, you know, the more you know a little bit about design, about coding, about marketing. So, what do you think is the new playbook for an entrepreneur?
Domain knowledge still matters. So, take that teacher. People in Silicon Valley, no one has teaching experience. That teacher has an edge because he spent like, you know, 15, 20 years teaching. He understands how to sell to other teachers. He understands how to market to other teachers. So, if you are someone who has domain experience that is unique, and the more niche it is, the better. But, to your point about what kind of skill set do you need as an entrepreneur, right? Being the hacker coder is not going to be as important. Actually, I made this argument on a podcast that went viral and was very controversial, and I said, 'Entrepreneurs that don't have coding experience might be at an advantage because coders will still kind of look at the code, and will still like think about the architecture, and they want to use the latest and greatest thing, and they're going to lose a lot of time and distraction just playing with things. Whereas, if you're an entrepreneur focused on sales and marketing and getting a product in market,'
Just KPIs, you're at an advantage. I think Paul Graham wrote about this, you know, 15 years ago. He called it resourcefulness. And the way I think about resourcefulness is never being blocked. Like, anytime you see a wall, you're like, I don't know how to open that door or that wall, but I'm going to bulldoze through it somehow. Being able to reach out to people, network, and find someone who's maybe an expert in this. Or figuring out how to use AI very well in order to deploy the AI's skills and knowledge to solve that problem is going to be important. The other thing is being plugged in and knowing what is on the edge of technology is very, very important. Also, being plugged into the culture, I think is important. That's how you get new ideas. That's how you can spot like a trend or a need. You can see what people are talking about, what people are interested in, what problems are they facing. Where I think you can gain quite an edge is understanding social media and influencer marketing and things like that.
Do you do any influencer marketing at Replit?
We relate to it. I am like the influencer of Replit, yeah. A lot of how people discover Replit is on podcasts that are always like on Joe Rogan, Diary of a CEO, and things like that. And we see the numbers actually go up when I go in these places. But we should have built like an influencer marketing machine earlier. Now we're building it. What you realize is that paid ads, Google, Meta, things like that, has become more of a tax than an actual channel. You want to play deep defense there as opposed to offense. But on influencer marketing, that's your opportunity actually to find a channel that could actually work for you.
Okay. Do you know Alex Hormozi?
How much did he make from his book? Like $100 million?
He just monetized his audience.
$100 million, right? And so, building the audience, then monetizing it is a very, very interesting approach as well. So, what does all of this mean? According to Amjad, vibe coding doesn't eliminate competition. It changes where the competition happens. In the past, having a technical advantage was often enough. Today, the winners are increasingly the people who understand a market better than anyone else. Domain expertise matters. Knowing your customer matters. Having good ideas matters. Being resourceful matters. Staying on top of new technology matters. Understanding culture, trends, and how people behave matters. And perhaps most importantly, distribution matters. Knowing how to use social media, build an audience, and get a product into people's hands become a massive competitive advantage. In other words, the bottleneck is knowing what to build, who to build it for, and how to get people to use it. Which raises a much bigger question. If anyone can now build software, what happens to the companies whose entire business is selling software?
Today, not everyone is into vibe coding. Not everyone has the time to do it. So, you still have a big window of time to like make something. And I go back to this domain knowledge. That teacher that built the teacher tools, his domain knowledge helped him build the initial product, but also helped him navigate the bureaucratic US education system in order to make any kind of sale. There are still moats in the world. Like Palantir, what is so special about Palantir? They know how to sell to the government. I don't think vibe coding takes away all moats, right? There you can still find moats, and they could be proprietary knowledge, proprietary technology, proprietary data, network effects. Like ChatGPT right now, although chatbots kind of commoditized, it's still the king.
