The Exchanges

Q Can I ask a weird one? Did you always know that you'd be successful? I think it's, it's sometimes an innate feeling. Sometimes it's not. Did you know that you'd be successful?

A Yeah, because there was a lot of sort of early successes. So that situation we were in My brother, my older brother and I, we found an arbitrage opportunity. The toys in the not-so-great part of town were actually much cheaper than the rich part of town where we went to school, and so we started selling toys at, like, very, like, preteen, young age. My mother would come pick us up in this, like, this beat-up old car, and we would put the, stash the toys in the trunk, and so we would sell them We would kind of tell people what we have during school because we don't, we didn't, we couldn't carry the toys, and then we'd tell them, meet us outside after school, and then we'd go to the trunk. My mom didn't know what we were doing, so we would sell these toys, and we also sold toys that weren't necessarily, like, Most savory. So we sold like BB guns and we sold all sorts of like, you know, non, uh, you know, so, so, you know, we, we, we did a lot of stuff like that early on to, to make money. And, um, and, you know, later on when I was a teenager, I also started, uh, my first software business. And so it, it just felt like I sort of had it in me. I sort of, I had the hunger. I, I knew I was, um, you know, I was bright. I did well in school, even without trying. And, and so, yeah, I, I sort of knew that I wanted to do something great. I also was attracted to greatness. Like, I really …

AI assessment note: “Yeah, because there was a lot of sort of early successes.”

Q The misfits scale. It's a great culture to have early, you know, you're 75 people. Now when you're 400 people, 500 people, does the misfit culture actually scale with process, with HR, with compliance, with approvals, with company wide culture? Can you scale and still have misfits?

A Yeah, um, yeah, I mean, I get, I guess I'm dealing with that right now, so I can get back to you in a few years. I would love to think you can, like, you know, I think there, there's like a, there's a way of, of, of, of doing that, but, you know, the history of companies show, like, you know, Wozniak, you know, was sort of like marginalized as, as Apple grew, and like, He's like a total misfit. Still is. Uh, but he stayed there for a long while. He was working on the Apple II, even, even when the Mac was the, was the thing, like, um, and he continued to contribute and, and, and things like that. So there are probably some ways of, of making it work. However, I will submit that, um, you can't build a company made entirely of misfits that, that scale.

AI assessment note: “you can't build a company made entirely of misfits that, that scale.”

Q When you look ahead, I see the green shoots. I see the creativity. I see the enablement. What, what does concern you?

A Yeah. So wealth inequality will grow insanely a lot. And how so? Because like people look at Instagram or WhatsApp as these outlier companies that were able to achieve massive valuations with like very little, with very little people. And I, I don't think it's going to be an outlier in the future. I think we're going to see a lot of people, a lot of companies, five, 10 people that are going to have massive impact. Probably be valued in the billions of dollars. And I think it's going to create a new crop of millionaires and billionaires, um, in Silicon Valley and elsewhere. Uh, and these people are going to have more and more power in the world. Um, and I think that will create more envy. I think humans will just continue to get richer But I think wealth inequality is concerning, not because it means that, you know, some people are getting sort of the short end of it. Sometimes that happens, but I don't think it happens in technology. I think what happens is just creates more envy because people think they look at, you know, rich people and they think, oh, you must have robbed someone to, to, to, to get there. Um, that's how politicians kind of frame it as well, right? And when you hear Elizabeth Warren and these kinds of people, uh, talk about rich people, uh, Um, they talk about as if you're, as if they're like thieves, right?

AI assessment note: “wealth inequality will grow insanely a lot.”

Q Three years. I saw a similar statement, which was GitHub CEO said that 40% of their code today for the whole company was, was, uh, AI generated. Where is that in 10 years? And like, is there an increasing speed of development? Do you see what I mean? Like, is the cadence of development increasing with time?

A So, uh, there are two ways in which large language models are gonna change software. So one is the AI-assisted developments. Uh, GitHub Copilot is kind of leading in the space. I think we're like a fast follower. Perhaps we're going to exceed them in, in the future, but we are, uh, Repli Ghostwriter is, is another sort of AI-assisted development. Actually, by the time the episode comes out, we would have come out with Ghostwriter Chat, which is, imagine ChatGPT that, um, embedded in your code, knows everything about your code, can answer questions about your code, really starts to feel like another pair programmer, human, that's sitting there and helping you. So that's one track of, uh, ways that it's gonna help us become better software developers, but there's another way which large language models can help us make better software and make more software. They are, in themselves, software. So, every large language model is a function, essentially. Like, you give it input, it takes output. And that function is programmable in English, what we call prompting. And you can make it do whatever you want it to, to do, right? So, you know, crawling a website and scraping a website and getting useful, useful information out of that is somewhat of a, you know, we have a lot of libraries and ways of doing it, but it's still a difficult task. You have to write a parser. It's at least a ma…

AI assessment note: “instead of spending weeks writing a parser, uh, you can just plug in”

Q uh, which is always dangerous. Never trust a VC who's an armchair economist. Terrible trades. Um, but like, you know, when you look at simple demand and supply theory, when you actually allow the world to create code, and when you create this access within developer time, you have an access of supply. Uh, will we see the price and the real wages of developers go down as a result?

A You know, Probably like the, I think there are developers today that are commanding insane, uh, amount of salaries while actually working very little. And like, it goes to what we talked about. I'd like the value of hard work and all of that. The idea that you can like go coast at big tech and still be like a millionaire, I think it's absurd. And I think that will go away. I think that's not just a, there's just the effect of an inefficient market, right? And perhaps a zero interest rate market. But, but I, I think that was an unnatural thing that happened. And so in that regard, yes, maybe the average, you know, developer salary would go down, but I think the, the most creative, um, you know, most hardworking developers will actually be able to earn a lot more.

AI assessment note: “yes, maybe the average, you know, developer salary would go down”

Q Amjad, if you could tell your twenty-four-year-old self one thing, coming to the U.S. with nothing but credit card debt, didn't sell enough of those BB guns, my friend, what would, what would it be that you would tell yourself?

A Not to sacrifice health. And maybe friendship a lot. Like, I think I was so driven that I, like I'm paying down a lot of sort of health debt and maybe reconnecting with friends now and, and things like that. I'm like, like, you know, knock on wood, totally healthy, but like there's some, you know, something just, you know, getting, getting fed and getting, uh, getting back into like a good, uh, Life habits and, and all of that has been, uh, has been a lot of work and, and, um, and so I've also neglected some, some relationships and, you know, I, I don't know. It's, it's almost, it's almost like, you know, the tweet recently by Mr. Wonderful. What's his name?

AI assessment note: “Not to sacrifice health. And maybe friendship a lot.”

Q Um, but with the, with the removal of other parts or with, with less engineers or with removal of other parts of the org, reducing headcount cost in those areas, are you able to justify sales reps moving down ACV categories or ranges because you don't have the cost elsewhere in the business?

A It depends. There's a lot of companies that now have self-serve enterprise, right? Uh, especially in AI, because there's so much, there's so much demand that you can sort of justify that. But let's say, you know, traditional SMBs still want, need some, some handholding. And it, you know, typically the, the, the cost of sale is higher because you have so much supporting staff. But if you can have a commercial, uh, AE that, that is, You know, you're willing to spend like an hour or two talking to a customer and onboarding them, and perhaps a few, few hours, um, you know, for the rest of the year supporting them, and the contract size is like 10,015 thousand dollars. I think, I think you're right. It's, it's possible that becomes more of a thing.

AI assessment note: “I think, I think you're right. It's, it's possible that becomes more of a thing.”

Q The misfits scale. It's a great culture to have early, you know, you're 75 people. Now when you're 400 people, 500 people, does the misfit culture actually scale with process, with HR, with compliance, with approvals, with company wide culture? Can you scale and still have misfits?

A Yeah, um, yeah, I mean, I get, I guess I'm dealing with that right now, so I can get back to you in a few years. I would love to think you can, like, you know, I think there, there's like a, there's a way of, of, of, of doing that, but, you know, the history of companies show, like, you know, Wozniak, you know, was sort of like marginalized as, as Apple grew, and like, He's like a total misfit. Still is. Uh, but he stayed there for a long while. He was working on the Apple II, even, even when the Mac was the, was the thing, like, um, and he continued to contribute and, and, and things like that. So there are probably some ways of, of making it work. However, I will submit that, um, you can't build a company made entirely of misfits that, that scale.

AI assessment note: “you can't build a company made entirely of misfits that, that scale.”

Q Token costs today, how do they change over the next three years? Is it like they get 50% cheaper or like 10 times cheaper?

A It's an interesting question because there's different ways to slice it. Um, you can look at the price of intelligence. The price of intelligence have gone down tremendously. It's really hard to quantify it, right? But like everyone's spending way more on opens than they were spending on GPT-IV, but they're getting way more intelligence There's way more productivity out of that. Pure, pure unit price is not going down as much as people expected it to go. Um, and I think part of it is, I think we're, why is that? I think part of it is we're living in a world where the true frontier models is actually just between GPT-V and, and, and Claude. Um, and, and so there's no, not a lot of pricing pressure. Now we're seeing Gemini catch up. We're seeing the open source models catch up. And so maybe as, as there gets more pricing pressure, we'll see, we'll see Token, unit token prices go down. Also, the other thing is, I don't think these, these companies are all that profitable, partly because, you know, the underlying, um, chips are not all that cheap, uh, and there isn't a lot of competition. We have sort of, at bottom, everyone's running an NVIDIA, and there's not much competition there, and NVIDIA has amazing margins, especially for a hardware company. What is it, 80%?

AI assessment note: “Pure, pure unit price is not going down as much as people expected it”

Q Token costs today, how do they change over the next three years? Is it like they get 50% cheaper or like 10 times cheaper?

A It's an interesting question because there's different ways to slice it. Um, you can look at the price of intelligence. The price of intelligence have gone down tremendously. It's really hard to quantify it, right? But like everyone's spending way more on opens than they were spending on GPT-IV, but they're getting way more intelligence There's way more productivity out of that. Pure, pure unit price is not going down as much as people expected it to go. Um, and I think part of it is, I think we're, why is that? I think part of it is we're living in a world where the true frontier models is actually just between GPT-V and, and, and Claude. Um, and, and so there's no, not a lot of pricing pressure. Now we're seeing Gemini catch up. We're seeing the open source models catch up. And so maybe as, as there gets more pricing pressure, we'll see, we'll see Token, unit token prices go down. Also, the other thing is, I don't think these, these companies are all that profitable, partly because, you know, the underlying, um, chips are not all that cheap, uh, and there isn't a lot of competition. We have sort of, at bottom, everyone's running an NVIDIA, and there's not much competition there, and NVIDIA has amazing margins, especially for a hardware company. What is it, 80%?

AI assessment note: “Pure, pure unit price is not going down as much as people expected”

Q How did you get over that, like, fear? And what would you advise me?

A I, I think taking massive risk is, is, is the key. It's like, if, if you really, if you really worried about mediocrity, then the best way to sort of run as far as possible from it is to do something so weird and so different that it's almost unthinkable. And so I guess in dating scenario, I think, um, you would sort of, um, maybe you would, uh, you know, not Attract all, you know, all the sort of like a vast majority of people would be actually, you know, turned off by this. But I think the people who would actually like you and would want to be with you for a long time and would actually match your sort of personality, you would attract them. And I'm actually a big fan of, of sort of being authentic and, you know, attracting allies and potentially rebelling people you don't like anyways, repelling them.

AI assessment note: “taking massive risk is, is, is the key”

Q Do you think you're creating a religion with Replit? And how do you self-identify with those that want to create a religion versus those that just want to be in a hot company?

A I think it's sort of in retrospect, you can sort of analyze these companies and say, I think it'd be like in the middle of it to be super intentional about creating a religion. It just comes off as cringe. I think maybe I'll let historians sort of decide that. But I think the main tenets is, um, you sort of have Somewhat of a, you know, Peter Tia calls it a secret, but some, somewhat of an insight into how the world works, or how the world should work, or where the world is headed, and so do you work to advance that. One of our biggest insights at Replit is that software is a superpower, and the superpower is actually sort of stuck in this ivory tower of, like, Silicon Valley elite, and we're trying to free this superpower and get more and more people, um, to experience it and make use of it, and we think that, you know, um, it's almost like, It's almost like the, at the time of the, like the printing press, right? It was the people who could read and write were the priesthood, right? And, ah, and, and then at some point, everyone in the world started doing it, and that changed our world. And we think software kind of follows the same trajectory. So that's sort of our main insight.

AI assessment note: “I think maybe I'll let historians sort of decide that.”

Q I, I heard from one of your investors, you were very creative in the way that you dealt with an employee that didn't scale. How did you deal with them in this situation?

A Well, I, I, uh, typically try to, uh, find places for them and in the company that there would be, uh, that would be still happy. Um, you obviously also want to, want to make sure that they say face and like the, the, the change doesn't look like an obvious demotion. Sometimes not always, but sometimes, uh, there are ways to find the intersection of where they would be happy and productive and where the org Really needs to be. And, and so I, I don't claim to have like a magical solution here. It's like one of the toughest problems to do as you're scaling a startup. I think you can also be too hasty instead of, uh, letting that person go or replacing them. And instead there are some ways to, Like find a place for them in the organization that would be happy for, for a longer period of time.

AI assessment note: “typically try to, uh, find places for them and in the company”

Q I, I heard from one of your investors, you were very creative in the way that you dealt with an employee that didn't scale. How did you deal with them in this situation?

A Well, I, I, uh, typically try to, uh, find places for them and in the company that there would be, uh, that would be still happy. Um, you obviously also want to, want to make sure that they say face and like the, the, the change doesn't look like an obvious demotion. Sometimes not always, but sometimes, uh, there are ways to find the intersection of where they would be happy and productive and where the org Really needs to be. And, and so I, I don't claim to have like a magical solution here. It's like one of the toughest problems to do as you're scaling a startup. I think you can also be too hasty instead of, uh, letting that person go or replacing them. And instead there are some ways to, Like find a place for them in the organization that would be happy for, for a longer period of time.

AI assessment note: “typically try to, uh, find places for them and in the company”