The Exchanges

Q market making because just the economic incentive is there. And so when something is at large scale, that's not as big of an issue, but I'm curious, What that was like in the beginning, like, were you guys doing the market making? Did you work with market making partners? Now, how do you incentivize market makers to participate? I'm just curious what the market making scale up has looked like.

A So there's actually, Two groups of contracts on, on markets, on, on caution, they behave very differently in the market making incentives are actually pretty different. So you have the long tail of markets, right? Like the ones like, well, one direction have a reunion or, you know, all those things. And they are actually very hard to price. And because there's not necessarily a lot of demand, we actually have to incentivize market makers, um, to, to come in and there's like liquidity incentives, all those things for them to come in. And I think it's actually how we think about how to build our moat long term is actually how do we get very, You know, sustainable solid liquidity in this long tail of markets. Yep. So we can get like, we have like, I think, 10,000, how do we get to 50, a 100,000 markets with still. But on the other side, you have the more classic, like, crypto sports, all of those guys, and, and on that side, it's actually a lot easier to market make, because you have very clear, proven demand, it's a lot easier to price. So the market making incentives on this side is actually, we don't pay them for it, we just rebate fees, but they have very, Very hard conditions to meet. They need to have uptime of certain amounts, spreads, and, and top of box size, and all of those things, um, because we see it more as like incentivizing stability of the book than it is incenti…

AI assessment note: “we don't pay them for it, we just rebate fees, but they have”

Q So why should you care about the code?

A Why should I care about the code? Um, you know, I care about correctness, I care about robustness, and I think, I know intellectually, I don't need to look at how the compiler unrolled this loop to verify its elegance and correctness, yet somehow I feel that way about code, and I'm not sure the code doesn't matter, but I've been trying to force myself to not care because I feel like I won't be like a self, Self-actualized software engineer in the future, if I'm too precious about that artifact, which used to be so central to me. Right now, writing Markdown files, like maybe that's fine. It feels somewhat like a local maximum, and maybe we'll just be like, oh, of course it's Markdown is how, you know, we work with machines. If you think about what a compiler does, um, there's this interesting mix of like formality and informality, and if you've used like Python versus Rust, sort of different ends of the spectrum, Now that you're not writing the code, I really wonder what that programming system should feel like and look like, and I don't mind chatting with Codex, that's fine. But I also think, you know, as you imagine, like all the tests that you care about, all the, like it showing you demos and mockups, and I wonder sort of what the future integrated development environment, for lack of a better term, will be in that world. So what I'm trying to do is force myself to not be em…

AI assessment note: “I care about correctness, I care about robustness”

Q What have you learned in the OpenAI board?

A Uh, a lot. Um, I mean, certainly the most interesting part is the AI research. Um, you know, I've never been affiliated with a true research lab before, and that's fascinating to me. Um, I, uh, it is very inspiring. I mean, it is, it's very easy to grow, not cynical, but like, you know, you can look at, you know, OpenAI, Google Anthropic, and say, like, who's, you know, whose model scores better on this leaderboard, to actually go in and see this company where every single researcher is trying to make safe AGI and not come out of this board means inspired is impossible. Like, it's amazing. Um, the other thing is, it's the first not-for-profit board I've been affiliated with, um, and, uh, that's really interesting as well, just because. It's a different thing, yeah. Well, and I mentioned the fiduciary duty is you have a duty to the mission. Yeah. And, uh, that is really clarifying and interesting as well, because when you're making decisions and you realize, you know, you have, your sole duty is to ensure that artificial general intelligence benefits humanity. Um, that's really different. It's really interesting. I've never had a fiduciary duty to a mission before. So that's really interesting to me because I take those duties really seriously and like reflecting in a board meeting and you're making a decision You think about it very differently, um, through that context. Um, an…

AI assessment note: “I've never had a fiduciary duty to a mission before. So that's really interesting”

Q um, uh, car and that license plate. Instead, you can just kind of much more quickly get it. And yet, this kind of stuff ends up super controversial from a, I don't want to say super controversial, but it ends up controversial from a privacy point of view. And so, I don't know, is, is that take too generous? Is there a steelman of the other side? Why is it?

A There is, I mean, I think you're right on the controversy. I would articulate it as like, if you're building a business That impacts millions of people's lives. It's gonna be controversy of some degree. Like, I, I'm sure people, I'm sure there's someone who hates Stripe. I don't know why they would, but I'm sure that person exists. Just like there's someone who hates Walmart. It's like they're trying to sell cheap groceries. Why do you hate, it's like, people hate, people hate every company. Um, maybe they hate us more. I don't know. Um, I, I think there's a few things that make it for the, for the steel man argument. One is you can see it. All right, so I bet you if we pulled up your iPhone, And we looked at the number of apps that you've given full-time local location services. It would shock both of us. And then if we looked at the number of data brokers who then leverage that data to sell you ads, we'd be really shocked. And I think if we rewind to 30 years ago and said, imagine these private companies tracked your location in real time and sold out to advertisers, we'd be like, that is unacceptable. But because we can't see it, we kind of let it go. Um, we have this perception of anonymity or being anonymous.

AI assessment note: “I think there's a few things that make it for the steel man argument.”

Q What does your day to day look like? And in particular, how are you managing by walking around? What virtual corridors are you wandering to just get a sense for what's going on on Microsoft? What are your customer engagements actually look like just for a normal day, not earnings or not a board meeting or something?

A Interestingly enough, um, my normal day is it, it's the two ends of it, right? Which is the, the customer stuff. So, There's not a day that I would say I'm not having, um, many of them are remote. I mean, there's Teams calls for me most of the day, at least two or three of them, uh, with some customer. It's sort of the most helpful way for me, um, uh, to stay most grounded, I would say. So I have at least one or two of those each day. And then, um, I'd say there is a lot of meeting time, um, uh, you know, Uh, you know, as a CEO, one of the things I've recognized is there are two types of meetings, right? One meeting is where I'm just supposed to convene and, you know, keep my mouth shut, because convening was the real thing, right? It is like, don't overperform, just sit and, because all the work would have either happened, uh, or will happen after, but, uh, so that's kind of one. And then the other meeting, which are the important meetings, where I do need to learn, or I need to make a decision, or communicate something. So meetings is another spot, uh, Then, I must say, it's kind of like all over, for me, Teams channels, right? I am lingering around Teams channels, uh, and they're most helpful. In fact, if anything, I learn the most there, I meet most people there. So, wandering the halls, I wish I can tell you that, you know, that is the form.

AI assessment note: “I am lingering around Teams channels, uh, and they're most helpful.”

Q Do you think we need to spend more time trying to discuss and meta edit the R&D system and incentive system that we have, um, because it just has a such a huge effect on people's quality of life?

A Well, yeah. Thanks for the question. I love to talk about this. So I think a lot about this and I think If your point of view is that we want more new medicines, like that would be a better outcome for the world. Yeah. And I am too. Then I think there's definitely many flaws with the current system. Strangely, most of the discussions I have about this are actually we have, they don't say it out loud, but there's enough new medicine. And what we really have a problem is affording it. Now, interesting fact. In the US, the most expensive healthcare system in the world, we spend 10 cents on the dollar on medicine. The other 90 cents go through everything else that medicine's trying to prevent. Go back to.

AI assessment note: “Well, yeah. Thanks for the question. I love to talk about this.”

Q Yeah. So that's one thing. You have to be somewhat hard-edged. What else?

A I think he works really hard. Um, I think that he compounds his own knowledge. I think he's super smart. Um, I think he understands that, that, you know, physics is the law. Everything else is a recommendation. But yeah, I don't know. At all, really. So I can only kind of speculate from the outside. Um, actually I think that one of the things that frustrates me somewhat about Isaacson's book is like, Isaacson is quite clearly a very sharp guy, right? And he spent a year and a half with Elon, and yet if you read his book, particularly like the last third or a quarter of it, which is talking about all the recent stuff, the Twitter stuff and so on, it's just, it's very clear that he doesn't get it. He doesn't understand why Elon is doing these things that seem completely crazy, right? But yet, He keeps, like, rolling sixes. Like, why is that the case? And my point of view on this, and I wrote a blog post about this at one point, was if I see someone who is pretty successful do a series of crazy things over and over again, and yet they work out every single time, either they have root access, To the simulation. Or their model of reality is better than mine in some key way that I don't understand. But I might be able to understand if I assumed that they made a sensible decision given knowledge that they must have but I don't have. What can you infer about this?

AI assessment note: “I think he works really hard. Um, I think that he compounds his own knowledge.”

Q Don't we also maybe get bailed out by AI productivity growth?

A Maybe, but do you really think that, like, 10% year-over-year productivity growth is going to, like, localize itself in the UK? No, but, but just, um, if you say it occurs in the United States, the United States is going to be like, oh, here's a big transfer payment for all your old people. I don't think so. Right. So it is certainly the case that AI seems likely to produce enormous additional wealth. Whether that wealth will find its way to alleviating the tax burden on the working class in, you know, the 200 countries on earth that don't have their own AI industry, seems somewhat unlikely to me. Whereas a pill that, like, I mean, actually audio's MPEC probably does this, slows down your aging by some factor, um, would be enormously important. No one could have seen this coming. No one, no one could have possibly anticipated that social security and other stuff would result in ever increasing dependency ratios. Um, and we're just fortunate that we got rich so fast that it took this long to bite, uh, it's a big problem. Yeah.

AI assessment note: “Whether that wealth will find its way to alleviating the tax burden... seems somewhat unlikely”

Q So is it because you're saying, like, the job is so pattern recognition oriented that you just need some time to build up the pattern recognition, and, like, is that what they're learning?

A Some people come, some people, like, Once in a while, people come in and like, you know, right out of the gate, you're like, wow, this person like sees it. You know, they see the ball really clearly. Yeah. But that is a very small fraction of the time. Most people get great over time, and they have to learn an industry. So when they come in, we will say, you know, you should cover, you're going to cover FinTech. And it just takes them a year just to understand FinTech, right? Then they have to see, you know, well, why is, XYZ stock trading at this multiple, and why is this stock trading at this multiple? What's the market saying? And yeah, I think it's like, the market is constantly giving you data points. Um, some of them are, you know, false signals, and some of them are good signals, and over the long term, they're all good signals. And the people who do great at this job, you have to have like some commercial sense, which I think is like, probably just, you're either born with it or not.

AI assessment note: “Most people get great over time, and they have to learn an industry.”

Q So you think the tech is very cool, but just fundamentally being an equity holder there is hard?

A I think tech is very cool. I just don't think it's progressing at the same rate it used to because a lot of the great entrepreneurs have left, um, because they felt like the environment for, Starting companies for, you know, driving new innovation that was going to make a lot of money. Like, that wasn't something that was, like, looked favorably upon by the government, so the best people left, a lot of them went to Singapore, and that, to me, like, I think that's a shame because, um, with the right government and with the right economic system, China should be, you know, uh, uh, the most important economy In the world, and it should be a technological leader, and I think the geopolitical tensions would be entirely different, um, and I have a ton of respect for, you know, what China did from the 19 seventies to twenty-twenty. Like, I think it is pretty much an economic miracle. But what we've seen the last five years is, like, governments matter, and, you know, it matters what your political system is, and it matters how the government intervenes, and the due process matters.

AI assessment note: “I think tech is very cool. I just don't think it's progressing”

Q for people. Like when you travel, you probably don't directly book the hotel yourself on the website. You know, you have an assessment or something like that, because choosing the hotel might be desirable for you, but actually filling out the web form is not a value add activity for you. And so, what is your vision for how agentic commerce happens specifically? Because it seems relevant to your interests.

A It's relevant to my interest. I mean, it will be extremely valuable, and it will be something that, Um, will be done a lot and might even be like a majority of commerce or like on, on the internet. I don't know. Um, but like, uh, I think my role in this is going to be like largely into infrastructure providing, right? Like I, I, um, uh, this is the, our job is to keep our, like all the minutes of business of Shopify extremely current. And so, um, um, what, um, we want to do is make sure that like, Everyone's plugged into the various chat, uh, systems, and there's an MCP connected to every Shopify store, and, uh, you know, global catalog that's really nicely presented, so it's, um, we're building software specifically for, um, the open AIs and the clouds and the, the, the, um, perplexities to make it really, really easy for these products to show up and reason about them, and, um, and so on, because we think it's really, really important that they can show up. So, like, there's an infrastructure angle in it. But, like, how is being used? So, I think the best way to predict the future in most cases is just, like, frankly, look at what rich people buy for themselves, right? Like, you know, Uber is a wonderful company. It's also extremely predictable because rich people had car service, and, you know, so it's figuring out a way to scale that in such a way from cost basis so that ev…

AI assessment note: “we're building software specifically for, um, the open AIs and the clouds”

Q Not in specifically, but I'm basically saying, do you make active decisions, or is it like totally formulaic?

A Well, let me tell you how we do it as a firm, and then I'll give it to the individual. How we do it as firm is we try to, we try to make it as mechanical as possible. We're trying to get out of the psychology of whatever's happening at that moment. So you try to, you try to define a process up front. But, but you do want to be, you do want to be discriminating, and so we have a magic box formula. Things like, you know, the quality of the, you know, are the founders still running the company? Quality of the founders, um, you know, are, you know, are they, are they beating their numbers? You know, what's the growth rate? What's the second derivative?

AI assessment note: “we try to make it as mechanical as possible”

Q online, there's, there's a new growth number. It used to be that we had some like pretty clear benchmarks for what it, what it means to be like a great looking SaaS company, you know, like the three X, three X, two X, two X year over year growth over four years. Obviously that growth curve does not apply to many AI companies. How do you know you're doing well?

A I mean, we do look at those revenue numbers, but, but our North star is not on the revenue. It's about the number of users that are, uh, are sticking with us. And, um, now we have a 120,000 are paying and it's a simple, simplest way to measure that. Um, what we're increasingly doing is to look at, like, what's the success for our users? And that's that you've built something that gets Usage. Or for the, like, design phase, it's just getting eyeballs, getting some kind of eyeballs, and then it's getting, having them have high retention numbers on their users. And we have, um, we have many apps that are like hundreds of thousands of users on their applications as well, or like visitors. Um, and just the number of our users that get to that stage is what I, what I, we're like optimised, starting to optimise for.

AI assessment note: “our North star is not on the revenue. It's about the number of users”

Q growth company from the bottom up. Folks are, are just kind of adopting the product self-serve and building a bunch of cool tools in it. And then it seems like Because you're on this accelerated growth path, you're also starting to go up market a lot faster than most businesses would in this time in their life cycle. Uh, is that right? And how are you guys thinking about that?

A Um, yeah, so we launched the team plan, um, a month ago, and we've already have more than four, I think, 5000, uh, people who paid more to get on the, on the team's plan where you collaborate, and that's done by a company, right? Um, and from the team's plan, then just in general, there's, I mean, hundreds or thousands of larger companies and enterprises that have seen, oh wow, I can get move much faster. I'm not bottlenecked by engineering. In, uh, to the same extent I was before. Um, and what we're doing is just to ensure they have a great experience and get their, their, their questions answered. Um, for now, that's everything we do, but of course we want to figure out how can like engineer some lovable team, uh, as we scaled like the, the customer facing engineering function out, out help in getting as much value as possible with using AI to, um, Move faster as a business and move like you had thousands of excellent engineers at the company.

AI assessment note: “we launched the team plan, um, a month ago, and we've already have”

Q Have you ever been tempted to raise for one of your projects?

A No, but a lot of VCs have been in my DMS and they want to invest and, and, uh, I think now, because now like I'm financially independent, I made a lot of money. Now I could do it because now I'm at a point where, um, like it sounds fun to play with other people's money, you know, like, but also in a serious way, it sounds fun to go like to, to a bigger thing. Um, I mean, Something you typically say after you've been in a week, a week in San Francisco, right? You start talking like this, but I'm not convinced yet, but I think now it will be fun for example. But the problem is if you're a founder and you raise money, you kind of need to go big or bust, right? That's kind of, it's hard to stay in between. You can't like make ten million dollars a year. You kind of need to make a hundred million dollars a year. You kind of need to become a unicorn. Um, that's difficult because, uh, the odds of that happening are, are not super high. Like it's, it's a few percentage maybe. So most people don't get there. And they do spend like five years or I don't know, seven years of their twenties on that. And that's something where if you would be an indie, you would have a ten million dollar a year company and you have ownership. It's amazing. You have so much money.

AI assessment note: “No, but a lot of VCs have been in my DMS”

Q So are like Phoenix and Austin, these places doing a good job of permits?

A Yeah, like if you look at, um, you know, um, places in Arizona, like the, the Tri-City area, Phoenix being one of them, um, you know, Scottsdale and the Tempe area there. You look at, um, the Tri-City area in, in, in, um, in Texas, um, near Austin, or you look at, you know, what's happening in Dallas and the, the Dallas Plano area. There's these pockets in, in the, actually, in fact, if you look at the country, You know, a lot of this, um, growth is happening in the south of the country, and that's been true for a couple of decades now, and they tend to be correlated, meaning if it's easy for me to build apartment units and to build just construction in general, it, it, it tends to be a bit easier to also get the licenses to open up a restaurant.

AI assessment note: “Yeah, like if you look at, um, you know, um, places in Arizona”

Q so again, it's if you end up stuck, um, uh, an emergency foil sleeping bag just so you don't die of exposure, but it folds up to it's like that size, and so you can just stick it in a day pack, and, uh, now you are somewhat more prepared for the, uh, the elements than you were before. Um, I don't know, do you have a favorite recent purchase?

A I, I like the, I, I just really like these, um, um, I mean, someone going just, like, to ridiculously further and, and, and, and, and to just, like, almost, like, products that just celebrate craftsmanship in a way, even in places where it's just, like, no one else does, like, it's like, I, I, I, weird example, but, like, I, I, you know, take all these vitamins in the morning, and, like, I've bought these, like, uh, CVS vitamin Pill cases at some point, just for travel, like for travel, and, and then I found on Shopify like this, like Japanese, uh, perfectly CNC-built, like unbelievably beautiful pill cases. Like, it's just like, it's, like, the entire website is just like a love letter to, um.

AI assessment note: “I found on Shopify like this, like Japanese, uh, perfectly CNC-built”

Q And how do you think Chris cracked it so early?

A So Chris is just, Chris always, Chris's entire life has been this pursuit of, it's just how, how he thinks. He's just born to do this, and it's, it's been, you know, it's in pursuit. He uses these terms. He uses one term. He says, uh, what nerds do on nights and weekends. Um, is one way to look at it. The second way to look at it is, um, uh, bad, uh, good ideas look like bad ideas. Um, and then his third most recent version of that is like internet cults. Um, it's like, if it, if it has like a thriving subreddit, then like something's going on. It's the other side of the, of the, of the people's negative emotion on this is the things that become movements early. Like, the internet enables movements.

AI assessment note: “He uses one term. He says, uh, what nerds do on nights and weekends.”

Q So does this lead to a political realignment?

A I believe it, I believe it does, yeah. So I, I think this is the big thing. I think Martin Gurry is the guy who has, You know, you guys published his book. I think he, he really nailed it, and I think his, his thesis in his book came out in I, I, I, I think a lot of people said either, wow, he predicted Trump, which, you know, is true to some extent, but that, like, that's not the big thing that he predicted, and then I think his prediction is in some ways so fundamental that it's easy to just kind of take it for granted and say, oh, of course, that's what's gonna happen, but it's like, it's actually so fundamentally important, I can't stop thinking about it, which is basically true transparency, true transparency, true free speech is just, is a fundamental solvent, uh, uh, uh, uh, uh, basically dissolving all centralized institutional authority. Um, and, and the reason for that is centralized institutional authority is never perfect, and, and it often has problems, and in fact, it often has very deep and severe problems, as we were just discussing, um, and the kind of show that a government agency or a big company could put on to claim that they're better than they are, uh, that would have worked under centralized media just simply collapses under conditions of, of, of, of true free, uh, peer-to-peer communication. Like, it just, there, there are just too many examples of too …

AI assessment note: “I believe it, I believe it does, yeah. So I, I think this”

Q in talent or something like that, is it that you are building your own software where other companies might have bought software like a Workday or a Greenhouse or something? Is it that they Are using the existing software you have better? Is the process that will be spreadsheets in a traditional company are built with software? How do you kind of use the software in these sorts of organizations?

A Yeah, like, there are sometimes, but we still use a lot of, like, the, the traditional vendors. Like, one pattern is, of course, elementifying everything, like, making the, the data explorable for you to be able to interact with it, like, who's in the pipeline, what worked, who does the best references, like, all of that, all of that works, so you can double down on that. But two, it's frequently things that you manually do, that a lot of the current, like, there is a gap between where, where, where the agents are today versus what you could do if you have the technical skill set. And a good example is, like, How do you scrape all the right profiles to be able to reach out to the right candidates? Um, so you, like, analyze whether it's, you know, how much I should want to say, but, but, uh, the, like, try to detect specific things that we know worked, so you bring that across to the, to the, to the, to the people. On go-to-market side, like, there's just so many things you can do with, with, with additional amplifiers. You know, it goes from Understanding what case studies are relevant and creating a good pre-read for you before you go to the meeting, through creating the AISDR experience that we spoke about, to creating an entire deck experience, so you have like a pre-populated deck with the right numbers that is customized to that customer, which you want still the person to…

AI assessment note: “we still use a lot of, like, the, the traditional vendors.”

Q all of the context you're working with Exists in the repo. Uh, it is in text. Uh, it's kind of neatly organized to be executed and read by humans, and so there's kind of a good bounce there. Um, the problem is that customer service agents are not, uh, of that character, and so how do you actually smush everything into a format where your AI agent can answer it?

A Yeah, we, uh, we spend a lot of time thinking about that to some degree. Uh, one of our engineers called almost like we're creating like a domain specific language for specifying customer experience. You know, like what is the mechanism of specifying it? We use this metaphor we call journeys, which is, you know, what is a customer journey end to end? And what does the agent need to be successful in that journey? What tools does it need to access? What information does it need to access? And you can, if you think about the capabilities of an agent like skills and a coding agent, you'll add, ah, different capabilities over time as the customer is talking to you. The key thing that's been a breakthrough that is probably not surprising to, like, the technologists listening to this, but has been a huge difference between those, like, crappy chatbots of four years ago is the reasoning capabilities. You know, I think the, you know, well, we had one client who had acquired three companies, and they had three identity systems, three CRM systems, three of everything, and so they had this big IT project where they were going to unify all those systems, but I was like, why don't you just have the agent, like, go in all three of them and just think, and they're like, well, what if there's duplicate data, what if the data conflicts? They're like, you know, that's going to, and I was like, we…

AI assessment note: “creating like a domain specific language for specifying customer experience. You know, like what is”

Q Um, okay, I have so many more things to, um, to go into kind of jumping around, but I like this. Um, how's the business evolved? So you're now, you said around five hundred million in ARR, uh, selling to both law Enforcement agencies and, uh, corporates. Just have there been interesting changes in how you monetize? Is it just a question of scaling up?

A Yeah, I mean, I'd say the, the biggest challenge is, you know, two, three years ago, we were single product, single customer. Like, we had our neighborhood business. It was growing 20, 30% year over year, but it was kind of operating, and law enforcement was, was going over there really fast. We had one product, and then maybe made a mistake, uh, I, I know, um, Uh, RJ from Rivian was here, uh, some story, like probably built too many products for too many customers really quickly. Uh, and in hardware, that's really expensive. Um, hardware tends to follow this J curve of like huge capex investment up front to get the thing going, and then you monetize and it actually winds up being.

AI assessment note: “two, three years ago, we were single product, single customer.”

Q there, uh, at Microsoft, uh, during the, uh, the thousand dot com bubble, and it really was, uh, you know, Microsoft's share price peaked in the late nineties, early 2000, and then didn't surpass it until 2016, I wanna say. What did it feel like in 1999? In particular, did you know you were in a bubble, or was it like, oh, this is the new, this time it's different?

A It's interesting, yeah. So, yeah, in fact, I remember, I think, you know, we probably became the largest market cap company in 2000. Uh, we crossed G, I remember that. Um, you know, we were capitalite, let's say, right? That time around, it was like, I guess I was more like Sam at that time, which is somebody else's capital was being spent. Quite honestly, when I look back at it, at that time too, the financial cycle aside, it was clear, the secular trend was clear that this is going to, because even the, by then, the, the business models were also emerging, right? Even for Microsoft, the biggest lesson at that time was, oh my God, like even our first order play, Uh, oh, we got to build a browser. We got to build a web server. Uh, we've got to, you know, have internet protocols everywhere. All, oh, we got, you know, we had a website builder inside of office with front page. We did all the obvious things, but we realized that just doing the obvious things didn't make sense. We needed to reinvent what we were doing, plus one of the new business models was clear. So in an interesting way, that cycle Um, kind of came out of nowhere. I mean, it came out of, you know, what was just, you know, whatever, rational exuberance or what have you, but, ah, the correction in some sense washed away a bunch of stuff, but I would say the ideas persisted.

AI assessment note: “at that time too, the financial cycle aside, it was clear, the secular trend”

Q all three from my cold dead hands. Where do you stand on the debate of do people have loyalty to, and there was also the revolt when they tried to take away four, oh, was it? That's right. And people were really attached to that model. Do people have loyalty to a model, or do they have loyalty to an AI brand, and how does this affect your business strategy?

A I think that in the consumer products, uh, this was the first time we saw that, right? When you changed models, they are not sort of uniform changes, uh, and they impact People differently, right? And, uh, personality is one such thing or style or what have you. And so it just sort of is a new dimension. So in other words, it's also an argument that, oh, wow, this is a new dimension of perhaps differentiation, right? Uh, people will, you know, it's sort of, there's the IQ side of it. There's the EQ side of it. Um, and then there is all these style points and maybe that's That's kind of one of the things that people will, uh, steer things towards. But long term for me, uh, I think you have to kind of make sure that the models are most capable for the hardest high value tasks. And then you continuously optimize after you have access to that for what the task at hand is, right? So as a, a product builder for us, you know, my, my thing is have the model drop, which is the most capable. But then what's in production, uh, is multiple models, and my favorite, like, new thing in GitHub, for example, is auto, which is, I want to keep, you know, people still obviously love Sonnet, whatever, they, they want to use it, but at the end of the day, I really want the model picker, and it just can't be a dumb model router. It has to basically have the intelligence to know that this task deserve…

AI assessment note: “So as a, a product builder for us, you know, my, my thing is”

Q Is that one app? Is that 10 different apps? Like, it's kind of interesting. If you think about the productivity suite that emerged, there was, you know, three Big apps in, um, Word, Excel, and PowerPoint, but it's interesting that that number was not one and was not 40, it was three, and so how do you think about this?

A I think that that's right. I mean, to me, it will be a few, I think, and in fact, the reductionist person, you know, in me says, man, they'll be the same things, except the job they do is going to be different, because I think, you know, a table, At least at the human level, right? Uh, because we can all talk about, like, what tools will agents use to communicate with each other, right? That's a different thing, right? Right now, for the RL loop, they are simulating our production environment, but they will ultimately be more efficient in creating their own production environments to kind of RL themselves. So, but let's just leave that aside. But in order to communicate with us, I feel like We have discovered some good things that we like. We like spreadsheets and tables, and we like documents in sort of linear form. We like inboxes or messaging tools. So these are like reasonable UIs, except the question I think you asked is, how does this thing have when it shows up in an IDE with like a set of changes? You have to help me more than just Say, okay, now here is a file, go to that file. Like, like that directed plan, not just to execute, but for me to do my workflow. Uh, like one of the things that we are experimenting with this mission control in GitHub Copilot is that, right, which is, well, the idea is you go have five, six different branches in which you fire off all these …

AI assessment note: “to me, it will be a few, I think... they'll be the same things”

Q What does the process look like for something like that?

A I actually started out, uh, my career as a financial service analyst, so I had looked at, you know, all the card networks, all the processors. It was pretty clear to me that the competitive set in merchant processing was, um, very mediocre at best, and that their technology was not conducive to most internet companies, um, and nor did they have the tech stack that would allow them to adjust To, you know, what was happening in terms of e-commerce. And, you know, look, at the end of the day, it was like, I think that there's going to be one, maybe two companies that actually can provide the technology for companies to enable e-commerce or any online transactions. And, um, you guys seem pretty smart. Um, so, uh, and, you know, we do a lot of work on managing teams and, um, You know, huge market, great management team, weak competitive set, uh, is like, you know, um, perfect recipe for making a lot of money.

AI assessment note: “huge market, great management team, weak competitive set, uh, is like”

Q So it seems like you've done well on this, uh, kind of being a person on the internet. How much of a partisan are you on the topic of The indie hacker approach, because obviously, most of tech doesn't run on the indie hacker way, where you're building multi-million dollar revenue businesses all by yourself. Um, do you think way more people should be doing the indie hacking thing?

A So, I mean, I think the internet makes you a partisan, right? So, I, I don't think I've been partisan much, but I think, um. Until you got an X. Yeah, well, X does help with it, but you have to understand, 2014, 14, when I entered kind of the startup world, There was barely any indie, there was no indie hacking, there was no bootstrapping almost, except Patrick McKenzie, you know, who you know well. You would, you would find investors, you would raise money, you would grow fast, you would hire a lot of people, you would get a big office, remote didn't exist, remote work. So, um, Compared to that, like what's possible now, there's different, there's alternatives possible now, and you can make a lot of money with this. I don't think you can become a billionaire. Maybe we will see that in the future, but.

AI assessment note: “I don't think I've been partisan much”

Q open DoorDash, it's kind of the same categories and things like that. And it's less personalized to me than if I took my DoorDash history and put it into, um, put it into an LLM. Isn't that, like, shouldn't we be somehow using the fact that LLMs are pretty good recommenders? Within products. It's not just DoorDash. It's every product I use. It feels like those recommender capabilities are underutilized.

A No, I think you're, I think you're definitely right that there's an opportunity here where there's almost like the traditional school of thought, which is to use the information that you have, right? And, and you build, um, the best personalized, um, models that you can. And, you know, one of the things that LLMs obviously does is it kind of throws efficiency out of the wall, right? Right. From the wall. And it kind of throws as much compute towards it And interestingly enough, one of the things that spits out, you know, when you put in enough tokens and have big enough context windows is you're right. It has actually better models because it's, I mean, it's just using much larger, you know, data sets.

AI assessment note: “I think you're definitely right that there's an opportunity here”

Q to create liquidity. Cause like in the stock exchange, you don't have to incentivize high frequency trading firms to create, you know, sub second liquidity. They are very excited to take on that project themselves and build the high speed interconnect between New York and Chicago to accomplish and everything like that. And so is this just the stage that prediction markets are at or is there something fundamentally different?

A I think this is what I was talking about, which is that this idea that you need, so So maybe finishing that sort of line of thought, and then like, I'll get to the, to the answer there. You now have a model where you need liquidity to be built on the fly, much faster, much more dynamically, right? And, and the market makers, the traditional Wall Street market makers are not geared up for that. It's not like they can spin up a new desk to price like politics or price culture in, in an hour, right? And so, but this is the part that gets really interesting, which is like, and this goes back to the foundational principles around prediction markets is like, A lot of these markets, the people that will price them the best may not actually be the experts or the authority figures that you usually would think about. It's actually Random people, like, you know, that live.

AI assessment note: “market makers, the traditional Wall Street market makers are not geared up for that”

Q Yeah. What else has made, what else has gotten worse from an offense perspective?

A Um, I mean, this is one that's like a, you know, I don't know if I have strong feelings on this, on this topic, but this, this concept that we, we appropriately hold local law enforcement to a very high standard, um, of accountability and audibility, and I think that's very good. Yes. The downside is like, we don't with criminals, and so as a citizen, you don't be the victim of a crime, you get really frustrated that they're not working hard enough, and it's just very, really that. It's that they don't have the tools, they don't have the data, or they're not legally allowed To get to the data. And the warrant system is a very, very good thing. It's like a very effective tool. But like, there's a debate of like, should law enforcement have more ability to solve crime faster? Um, and the example of the, the framework that we use is, I don't know how you land this, is that the severity of a crime should be commiserate with the sophistication of technology. And I'll give you an example. Facial recognition. Hot topic. There are thousands of cities in America that have banned law enforcement from using facial recognition. John, facial recognition is not bad. That's the technology. It's not good either. It's just technology. I think a way more effective measure would be to say, hey, look, facial rec has its pros and cons. You can't use it for shoplifting, but for homicides, uh, crimes…

AI assessment note: “thousands of cities in America that have banned law enforcement from using facial recognition”