From the moment I was a teenager, I was making money just because I was good at coding. I'm like, 'Wow, this should be available to more people.' Right now, every day I hear of a user that's making a million dollars on Replit, right? Yesterday, there was someone tweeted out that they were making $8 million. And every time someone is shipping something and they're saying, 'Oh, this is working. Yay.' And I'm like, 'Can you tell me what's not working?' I think we're headed into a post-prompting world. And the next phase is about...
Let's dive right into it. Okay. Amjad, you know, you came to the United States when you were 23 with a bunch of credit card debt and no job, I think, right?
Um, no, I had an offer from Codecademy at the time. Yeah.
Oh, you did have an offer? Yes. Okay. Okay. Well, but no job. So, you know, it was technically accurate. One of the things that you know I think is super hard at the minus one stage and frankly even at the early stages is everyone talks in Silicon Valley about how persistence, resilience, grit is a feature, right? But there's a fine line between being incredibly kind of like stubborn and also kind of like sometimes being like this idea is just not that good. So how have you thought about that? It's kind of taken you 9 years. I mean everybody thinks about Replit is an incredibly successful company but it's taken you 9 years to get there.
Well, more than that, right? Like when I came to the United States, Replit was already kind of a side project. It was an open source side project. So, you know, 20 years. But the story on the credit card debt is interesting. So I had this job offer from Codecademy, but I didn't have any money to get to the United States. On my way out of Jordan, I had an O-1 visa. No one had heard of O-1 visas at the time. Right now they're so easy, but at the time it was hard. The airport, they stopped me. They're like, 'This visa is like, what is this visa? Is some kind of scam?' I was like, 'No, it's like a work visa.' They're like, 'No, you have to buy a return ticket because it's not an immigrant visa.' And they made me buy that. So, that was all my money. And I got there and the founders of Codecademy were like 20 years old and I was their first employee. And so I got there, I didn't have any money and they were out in Silicon Valley fundraising. So, I'm here in New York literally on my own, don't have any money. I had this credit card that I started back in Jordan and I understood how credit card works. I thought I had some money in it. I thought it would work like a debit card. And so the more I swiped it, I saw that I was looking at the balance. I said that the negative is giving money to me. Yeah. And so it got so bad I think I racked up like $15,000, $20,000 and my father calls me one day and he's like the police is looking for you back in Jordan. It's like why? It's like you defaulted on your debt. And then eventually I paid it off. But in terms of just the idea, I mean part of it could be insane. Like you could definitely be insane and be going after an idea for 15 years and not succeed. I mean there's so many stories of entrepreneurs. Maybe some of them eventually succeed. Maybe some of them just never succeed. But I think the thing that really drove me is I felt like I saw something about the future. And that is I saw what programming and being good at coding did to my life, to our life. I grew up from a very humble background and from the moment I was a teenager I was making money on my own just because I was good at coding. I'm like wow this opportunity should be available to more people. So the thing that always motivated Replit is how do you make programming more accessible and if you make programming more accessible that incredibly valuable thing because people can make money, can get jobs, and so surely you can find a way to monetize it, get value out of it. I just didn't know how and that took a lot of different paths. And the thing about Replit even 10 years ago we've always had success stories they were minor but over time and right now every day I hear of a user that's making a million dollars on Replit, right? Yesterday there was someone that tweeted out that they were making $8 million solo entrepreneur, which is amazing. But even back then, it's like someone was like, 'Oh, I learned to code a little bit on Replit. I'm like a fitness instructor and then I figured out how to do an iOS app and now I'm making like $1,000 in the app store.' Like, whoa, that's amazing. So, how can we create a million people like that? And so there was always something that showed us that this could be a big idea and that kept us moving. The thing that really excites us, you know, the charts up into the right, they're amazing when we talk to users and when you're building something that enables people, that's the best feeling. I mean, you two know it, right? Like you know the returns I'm sure are amazing, but when you see a founder that came out of here that is crushing it, that's the thing that makes it all worth it.
So I actually want to check in with Haya. So you know you were a designer. I think he asked you to design the logo for Replit. And then you decided to join him full-time on this journey. I guess like way back then I'm guessing he didn't have the messiah complex then but way back then what did you see in Replit? Like why did you decide to do that?
Yeah. So I had a job and Amjad came to me. He didn't have a job. So he was working on Replit he was a student and he was like I need a logo and I need a UI and I was like for what and he says I have this side project going on it's called Replit. No idea what read eval print loop even means. I was like okay what is Replit even mean. So we have to ask many questions as a designer we can be a little bit too much but it's part of the job. So he says well it's a ripple. I was like I don't know but I can imagine stuff I can make up stuff. So I imagined like a water ripple and I created a logo for that but I didn't know what I was doing. So I created the UI and he pushed the UI of course he changed the UI and then he came back to the same UI that I developed for him. It's the designer engineer relationship. You know, they think they know better, but they were like, 'Oh, this actually looks good.' Like, 'Oh, you mean exactly like this how it looked like before.' So fast forward my story with Replit. The reason why I started Replit with Amjad because I wanted to find a job.
Okay. Should we move on to AI? I mean this is a weird question to ask you guys but you know it feels like a lot of the momentum is swinging back to local development and as someone building like cloud native prompt to deploy sandboxes like how do you think about the resurgence of local AI tools?
Well I think there is the question of like the absolute numbers versus the share. I think in absolute numbers all of coding is growing whether it's cloud whether it's local. I do think that there's some fundamental limitations to local. So for example there was a supply chain attack in the JavaScript ecosystem just yesterday and it's pretty bad. Like if you installed a very massive one of the most popular libraries that does requests in JavaScript it would take over your computer. Like literally remote takeover of computer. It was insane. Yeah. So if you're in Replit, you're not exposed at all because not only is it a cloud environment that is not connected to your local computer. We also block post install scripts. We also have a minimum age limit for a package. So we don't allow packages that were just published. And it was caught in a few hours, but even a few hours many millions of people installed it. And so I think things shift, hype comes and goes. And at some point Replit Agent was the only agent on the market September 2024 and now we started an entire revolution. We're proud of that. At some point we're going to do something fundamentally new no one else did that will start again. We feel very proud that we've always been on the cutting edge of what's possible. At the same time if you believe in something you just do what you think is right. And for example, the enterprise side of the business right now is taking off primarily because of the security and reliability features because enterprises are like I don't want people installing open claw and all this stuff on their computers. I want these cloud environments that are safe and secure and been battle tested for many years.
I was going to say, I mean, I love our clawed overlords, but this local development makes zero sense. Like, it actually is so much easier to do everything in the cloud remotely managed in a way that's secure, but also autoscales. Like, I literally walk around San Francisco and, you know, whatever Uber I'm in with my laptop open, so the thing keeps on running. It makes zero sense, right? So I'm hoping that you know we come to our senses about that. Here's the question. So one thing that as a founder today is super disorienting is the pace of change. Like you are both trying to build something on top of foundations and abstractions that are incredibly powerful but constantly shifting in a way that they're both enabling more things but then also like kind of taking away what you are building and kind of like swallowing them into themselves. So how do you deal with that at Replit? Because obviously you're kind of trying to be at the forefront and you know the agent stuff is a good example but at the same time some of what you're building is getting integrated into the platforms themselves. So how do you almost visualize that race in your own head in a way that makes you confident about kind of where you want to go?
Yeah, we were talking about this actually earlier but I'm going to say two things that are contradictory and you guys figure out how to cohere them. Two truths, one lie. Okay. Yeah. Do you remember when Marc Andreessen went viral for quote unquote maxing and saying that he doesn't introspect on Twitter the other day and that was like super controversial and I think there's like deep wisdom in that and that as a founder you have to kind of be stupidly executing on a day-to-day basis and the question is is the company growing? Are you adding customers? Are you selling your product? And that is true for us every single day. And if not, if there's something wrong, just talk to customers. They're going to tell you it is competition. They're going to tell you it is something else. And handle these problems locally. And then if you do that enough days in a row, you're going to end up in a much better place, right? So, in some sense, you want to shut off the entire world and you want to focus on your customers and people using your product, right? That's number one. Number two, that is totally opposite is every now and then try to predict where the future's headed and try to predict your place in it. Right? Like I think actually predicting the future is much easier than most people think. Like my TED AI talk 2023 I came up and said software is going to be solved in the next couple of years. Here's what software development looks like. And I sketched out like vibe coding for the next two years. We actually just completed the road map that I sketched out like one of the last elements was parallel agents how you can put in a prompt and it becomes four different tasks. For example, Replit now I think it's the best parallel agent on the market. You know that was like looking forward like three years and so we have some deep beliefs about where the future is headed and so we're trying to position Replit in a way strategically making few key bets that will pay off in the next 6 to 12 months based on where we think the world is headed but on a day-to-day basis I think if you're consumed by these questions you're not going to be able to execute.
To kind of perhaps echo this point back to everyone here who's an early stage founder. I actually think that this is one of the key contradictory things you have to become good at in today's day and age which is that how do you find kind of like north stars and kind of like essential philosophical groundings that you feel confident about and also kind of ignore the day-to-day that you will hear around you but also not ignore it to the point that you are actually ignoring the good information. So it's actually just but I don't think anybody's figured this out. So I think if it makes you feel better, I think that this is a problem in today's day and age that all of us have to figure out as founders. And I think it's a really hard problem to figure out.
So if you're holding both states in your head and you've just accomplished what you laid out in 2023, what have you laid out for the next three years?
I don't know if I want to say it here. These guys are going to compete with us. You just tell us one thing. I think we're headed into a post-prompting world. I think prompting, you know the way technology moves is like things get layered right but I think we've reached the apex of what prompting is and what it could do and the next phase is about two different things. One is new interfaces for interacting with AI. And the second is fully autonomous agents that can push and set a pull that can actually operate in the background, understand you or your organization about how they work and do proactive work that is useful. And we're building in both these directions.
I'm a little confused about the interface because it will still be web-based.
I think that the multimodal capabilities of models are very compelling. Like in the movie Her like how when he's installing the operating system, he's sitting in front of his computer, he's talking to his computer. I think we're almost there in a moment where you can sit in front of the screen and maybe even turn on your webcam and maybe there's a touch component there and it's very natural sort of like you're dueling. You know that we just launched our canvas interface you are like moving your hands you're talking it was like do this here do that here. It kind of looks like Iron Man did you see that meme Iron Man vibe coding. And so I think we're headed to that world and maybe part of it is going to be VR or AR but we're headed to that world where communicating with machines just feels more natural like more like how I talk to you.
Or Minority Report but I don't like to give that example because it's like so dystopian but yes but that interface is really cool.
Yeah it's interesting both of the kind of use cases that you laid out. I was just actually literally vibe coding while I was building SPC's app kind of upstairs right before this and I was just trying to be like move this thing over here, shorten this, kind of like rearrange the layout, make it more compact. And then the other thing I was running was trying to just again automate the set of things that it's supposed to do every morning for me. But I was thinking in my head, I have done the same thing the last two weeks, like why is it not predicting what it's supposed to do? So I think both of those, which is inferred use based on my usage patterns, and then also the UI stuff can be so much quicker than what it is right now. The prompting back and forth is so slow. Yeah.
Well, you should check out Agent 4 the canvas. There's some fundamental way in which the big labs are operating that is limited. I think in trying to build that AGI, they're not really building systems that are super user friendly. Like you know Claude Code work is amazing and code is amazing but everything is based on this like prompt something happens and maybe you can schedule it so that it happens every day but it is not a fully integrated system that understands. And this is the opportunity for everyone here is like building amazing user experiences because I don't think...
I agree I think the hardest thing or my biggest user issue and my biggest user complaint is how do I load all the context that I want to share to be able to build something as powerful as I want it to be which is like going back to what you were saying like why hasn't it learned from my usage patterns why isn't it retaining context.
Yeah so that's where we're thinking about like the pull and push information and what would that look like. So right now it's a lot of the user is telling the agents like do this do that. You don't understand me like no no no like I meant that did you forget about XYZ we spoke about it multiple times. I mean and at this point maybe you need to understand my taste and I build multiple projects and I need it the same way so like if you're working with a real engineer or real designer at this point they need to understand each other at some point at some level this is what we're trying to do. One of the coolest things in Agent 4 is that the entire project context is in one place and instead of usually previously you would build a website one place a web app mobile app in another place Xcode whatever. So literally on the canvas you see your mobile app, you see your web app, you see your back end, you see everything and now you want to generate a slide deck to pitch investors. It's like generate a slide deck you know because it has entire project context in one place. You can do that and again those kind of interfaces they're not prominent because these big companies are just focused on the model being as powerful as possible. And so that leaves a lot of white space for entrepreneurs to invest.
So what is the most powerful thing you have built using Agent 4?
You know, our users at this point are just way better than me at building things. I'll tell you one thing that really blown my mind recently. I did a shoulder surgery and had a physical therapist and she always had the... she kind of specializes in fascia therapy. Not many people know how to do it has this trigger point component. And she tracks your progress every time you see her like lift up your hand what is your range of motion things like that. She always talked about this idea of like I want to be able to track that. I want to be able to take a 3D scan of your body and allow you at home to kind of do these exercises, give you feedback on them, track your movement, all of that. You know, I would tell her to check out Replit every now and then. And then I was like, most of the time it didn't work out. But like in the past few months, they actually built an amazing app where as part of the onboarding, you take a few pictures of yourself, it creates a 3D... Wow, that's amazing. 3D scan of your body. And they were working with offshore developers and they created like a mess for them. And now they restarted everything on Replit and it's just working. And it's like a healthcare app but it's like a delightful user experience. And I didn't know that Replit is good at doing 3D models and 3D experiences. And these are the kind of things that weren't built pre-vibe coding because the market cap for something like that I don't think is going to be...
That's the question I have for you like who is Replit's user is it like the creator the individual user is it the enterprise I understand it's both but I'm kind of curious about can you build a business model without enterprise because every other competitor or so-called competitor you might have there whether it's like Claude or Cursor or all of these other people like the main revenue even Cognition the main revenue source is enterprise so can you build a revenue model from this creative everyday user?
Yeah. I think so. I mean our revenue is now still dominated by individual entrepreneurs and small businesses. Our enterprise revenue is growing fast. I'll tell you some of the enterprise use cases but by the way I don't think these personas are that different in that I think what Replit unlocks today is I am a domain expert. I'm not a programmer. I know how to create value in the world. But previously I didn't have the tools to do that. So outside the enterprise, it is this doctor that has deep domain knowledge in a very niche subject where some sub population of the world are interested in an app for that. She can create that that previously wasn't created because if you're tech bro in Silicon Valley, you're not interested in that, right? So now that gets created. In the enterprise. For example, one of our customers, Zillow, there's like a product manager that was managing the real estate agent team that responsible for lead routing and managing the real estate agents to kind of be able to close deals. And he always had ideas on how to increase conversions, but didn't have the engineering resources to do it. He was able to do it and he created millions of dollars of value of the company and became super popular at the company was sitting on board meetings and hanging out with the founders. That's entrepreneurship too except it's inside the enterprise. So I think it's really the same type of user. The challenge for AI in general broadly today is that the concept of ICP is actually not very useful. Yeah. Because it's like oh I'm going after the product manager that went to school to do X Y and Z. Instead we're starting to think about traits like what are the traits of the Replit user and like resourcefulness. So things PG used to write about you know that the hacker mentality all of that is the kind of user that we're going after. So for most part people like to think about persona and not to think about persona is very difficult because you need to detach yourself from these personalities and start to see common traits between all of them and see and notice behaviors and what makes all of these people across all of their domain and expertise the behaviors are all the same which is I want to be heard. I want to do something. I notice a problem and I want to do something about this problem now. And these people are very stubborn and because of all of that too, they want to win and they want to prove that they're right, they're not wrong. So whether they're designers, they're PMs, they are small medium business owners, they are an enterprise, they all have these traits and qualities and it's very similar from the consumer to enterprise.
I'm curious I mean you more than you and Haya have more than almost anyone kind of a front row window into the evolution of the models and especially kind of the accelerating progress and capability of these models over the last three or four months right? And I think that you're laying out a version of the world which I subscribe to as well which is that these models have seen particularly kind of Claude has seen a tremendous rate of increase in terms of capabilities starting whatever December, January and right now in some ways we are in peak love with Claude Code because these models they feel like you want to be raw you want to be close to them and it gives you a sense of like wow between Claude Code, Claude Code Work and so on there's a tremendous amount of capability but pretty soon you're like well I kind of don't want five terminal windows running I want something that's much more managed I want different UI more capabilities right? I guess the first question is what do you think changed in December I'm curious to get your analysis of like what changed in the model capability and number two how quickly are you seeing users get in some ways frustrated with the minimal UI of these raw models and wanting more?
I think there are different phases of model improvements. You know say 2021 or like 2020 to 2023. It was a pre-training era where you just put in...
Bunch of crap really, any kind of data didn't matter and you just have as much parameters as possible and you grow this model and it just keeps getting better. We have the scaling laws and we thought we could... the simple word, more data, more parameters, everything was better.
Yeah. But then we hit a plateau, and the plateau for a bunch of reasons. One is like, you know, there wasn't a lot of data left in the world. Another is like the training infrastructure got like a little too complicated for growing these models a lot more. And so we entered the sort of RL and thinking era. So like say 2024, 25, with you know starting with o1 and DeepSeek and all that, you had this moment of time where the training regime changed and we entered this moment of you're doing reinforcement learning and these models are getting more autonomous because not only the raw intelligence is improving but they're staying coherent longer over time. I think the big unlock that happened just in November, December, I think the coherence... it just became almost indefinite.
Like Agent 3 was kind of the most autonomous agent in September 2025 and we struggled to make it run for an hour. But right now if you have enough evaluation in the middle, if you have computer use or unit tests or whatever, really the models kind of run indefinitely. The other thing is like models also gained some self-consciousness almost. I think that's too much of a loaded word but they started training on their own context window. So they actually also got really good at compression and really good at taking context and storing it into files and being able to search across these files and recollect the context based on the next steps that you're going to do. And all these things culminating together created these different models. The other big thing, at least this is the rumor, is that it looks like the big labs have figured out how to scale pre-training again.
Yeah. So I think we're going to enter another phase of scaling where we're going to see bigger and bigger models. So Opus is like a really big model. And I think they're going to get even bigger. And I think the reason token prices are not going down is because the models are getting bigger. But it's sort of ebbs and flows and you need to understand what stage we're sort of in. So I expect the real capabilities to continue improving in the near future. That being said, I still feel that there's a lot of white space between the AGI providers and what the users want and the kind of thing that is actually fun and exciting to use. That doesn't change.
100%. Like that's human and humans don't change all that much. We're visual creatures. We always want to have visual feedback as opposed to like just pure terminal.
Yeah. And I strongly agree and I think that you know if you take the other side of that, if you're the kind of founder who believes that in an AGI world there will not be that much for humans to do well, then my real question is why are you starting a company, right?
Go shop. Or go learn how to surf I guess. Yeah.
That's what I'm doing. So you could believe that.
Okay, perhaps Richie should move on to company building beliefs. Yeah. I mean, Replit has a pretty fascinating internal company culture. Like, how many of you have heard about it? Come on, guys. Well, this is your opportunity to learn. So, one of Replit's core values is seek pain. I cannot imagine putting up on my walls, seek pain. But like where does that come from? And like do you scare people away?
Yeah. So I'm going to answer because this is mine actually. I created it. Seek pain comes from curiosity when something is working or not working and going deep, deep and trying to understand what the hell is going on. It's really difficult to detach yourself from the work and not feel that it's personal. It's not personal. It's not about you. It's just work. And this is very difficult for people to go to users or go to actually people and ask them, hey can you use this thing and or like can you let me know how does this feel to you and they tell you it sucks and like you don't feel, you know, like you're now demotivated. So it's such a skill to master that skill. So it's just like learning that right. So, and then like that's there's the growth into it and there's like a lot in seek pain. Seek pain in hard conversation. Seek pain in understanding like what's going on in the product. Seek pain is how can I get better in XYZ. So all of that is a principle that I take to heart and I wanted to bring it to Replit and every time someone is shipping something and they're celebrating and it feels like a bubble that they're celebrating something and they're saying, 'Oh, this is working. Yay.' And I'm like, 'Can you tell me what's not working?' So anytime anyone's celebrating in the company, I will be like, 'Why are you happy? What sucks about this?'
What sucks? Can someone tell me what sucks about it? Because if I know I would be like more fulfilled to be honest.
I mean Richie you say you don't want to put this on the wall but that describes you completely. It does like every time we have a victory lap about SPC you kind of smack me down with like the 10 things that are not yet great about it.
Or I mean you could be better.
Could be better. Exactly.
