He Hacked Finance And Is Now Building An AI CEO - EP 63 Pedro Franceschi

Today's guest is a coding and entrepreneurial prodigy. He is Pedro Franceschi, the CEO and co-founder of Brex. The headlines on his life story are quite remarkable. He grew up in Brazil, taught himself to code as an 8 or 9-year-old. By the time he was a teenager, he was, he was hacking the iPhone. He was getting legal notices. From Apple asking him to stop hacking the iPhone. He was starting a number of businesses where he made hundreds of thousands of dollars while all the rest of his compatriots were on the beach in Rio de Janeiro or playing soccer. Pedro was coding away, creating these, these companies. He has some, some very deep sadness in his childhood. His father got cancer and passed away when Pedro was just 8. That left his mother to raise Pedro and his sibling as a single mother. She had some unconventional approaches to raising Pedro, which we get into in the show. Obviously we get into Brex, which has been this just superstar of modern financial tech. They were just acquired by Capital One for $5. $1.5 billion. We speak about this on the episode. We talk about Pedro's really open about the ups and downs running this business, his own personal ups and downs, struggles with mental health. He writes about these things. He talks about them in the show. Um, just, he's 29 years old, so this dude has lived quite a, uh, quite a juggernaut of, of a life so far. And. These days he's using AI agents in some what I find imaginative and comical and inspirational ways. He's got a team of AI agents running his life. We, we speak about all that. And so I should note Brex is the flagship sponsor of this podcast and our video series. As such, we are deeply journalistically conflicted in this episode. Nonetheless, I think it is, uh, it's a good listen and he has a fascinating story to tell. This is Ashlee Vance. You're listening to the Core Memory Podcast, and off we go. Pedro, thank you so much for coming in.

Thanks for having me.

You are— you're royalty in these parts. Um, you guys have been such amazing supporters of everything.

Oh, thank you.

What he's been doing. So very glad to help.

We're big fans.

I feel extra pressure, uh, doing this one. You, uh, at the end of last year you got married, you just were acquired for $5.1 billion, and now you're on the Core Memory Podcast. This It's all happening.

It's all happening. It's really all happening.

I wanted to, I wanted to, for people to get to know you a little bit. I mean, look, I cover, I cover all kinds of inventors, startup founders, you know, and so some stories I see repeating a little bit and there's, there's echoes of that in some of your biography. I'm just gonna read off a little list here for people, super quick. And then dive into a couple of these things. But you know, if you throw your name into Claude or ChatGPT and ask for like a quick, quick background, or, you know, it's like ages 8 to 9, teaches himself to code entirely via Google. No formal, formal training. Age 11, begins jailbreaking iPhones and iPods. Age 12, you know, starts creating apps. You're selling apps to, to friends at school. Age 15 to 16, Well, then this is kind of when you start your journey heading towards Brex one day. Yeah. Meeting, meeting your friend, um, Enrique. But, but, you know, so some of this is familiar. I mean, it reminded me a lot of George Hotz, who I've written about.

Oh, in OA?

Yeah. He's like this teenage hacker. He was, he was busted.

We used to be IRC buddies back in the day.

Okay. Yeah. You know, he was, he was getting into the PlayStations, but I mean, like, even among these people I've covered, like quite exceptional, especially running, you know, money-making businesses as a kid. So, and there's some of this in the things that I read, but I mean, tell me, were you just always, you hit upon a computer and it just, it just makes sense to you?

Yeah. So I think the biggest luck of my life was I found out what I love doing when I was 8 or 9. And most people don't have that luck, but I did. And my— I grew up in Brazil. I'm from Rio originally. And growing up, I lost my dad when I was 8. And he was a big nerd and used to spend a lot of time on the computer. We had a computer at home, which was relatively rare back in Brazil. And that somehow was something that was around my life since the beginning. And I would say when I was 8 or 9, I just had this very deep curiosity to understand how the computer worked. And then I opened it up one day and I saw there wasn't anything interesting inside the computer. So I thought, well, there's something else going on here that makes this actually be so magical and so amazing. And I realized it was software. And then as a kid in Brazil, I wasn't really exposed to software engineering in any capacity. But the internet existed and the internet's this amazing thing that, you know, allowed people like me to learn something that I would never be able to learn otherwise. So I started Googling how to code and started learning C and C++ back when I was 9. And learning that and English at the same time made it a little bit more challenging. C isn't a particularly friendly language, but it's all the resources that I found back then in, I guess, 2004 or 2005. So, that was sort of the backdrop of how I started coding. And what I really found remarkable about computers was you just had this ability of taking whatever you wanted with very little sort of permission or need to go do something. And then around the time I was 12, I became a little bit of a better engineer and it was around the time that Apple launched the iPhone. And if you remember back in 2008 or '09, the iPhone only worked with AT&T here in the US. So there was this whole thing of when you wanted to use the iPhone abroad, you had to jailbreak it to get it working in other countries. You would jailbreak it and then do the carrier unlock, but jailbreaking was the hardest part. So I was using a lot of the tools that existed online to make that happen. And then a new iPhone came out, the iPhone 3G. I ended up finding a way to jailbreak it. That was one of the first ones back then in 2009. And I got sort of known in this like iPhone, jailbreaking community. And then weirdly enough, that's actually how I got— I started working. So I used to, you know, a reporter in Brazil realized that there was this kid doing, you know, doing, you know, jailbreaking things online.

I mean, you were like a celebrity.

I guess I wasn't really known back then, but because I used to just publish things online with my handle, which was PH. My name is Pedro Henrique. That's why PH, like the chemistry PH. And somehow this reporter realized I was Brazilian. I think it was because of my IP address in this IRC forum. And he reached out and said, "Hey, there's a Brazilian involved in this iPhone jailbreaking community." And then he asked, "How old are you? I want to know more about you." And in my head I thought, "Well, if I tell people I'm 12, they're not going to take me seriously." So I told people I was 14. And in my head it made a huge difference. So, and then right around that time there was sort of in the Brazilian press and a company in Brazil that was building that made mobile apps reached out to me and I ended up going to work there right around the time I was 12, 11 or 12. And at first I told my mom, "Hey, this company wants to hire me." And she said, "No way, what are you talking about? You're 12, no freaking way." But then I convinced her to come interview with me. So she went there and I started working there part-time. And it was great because it was the first time that I actually met professional software engineers. Like what does it look like to really be sort of world-class? And my first boss, Ben Jackson, was fantastic. And, you know, he worked at New York Times after building their mobile app. And, you know, he went to Penn. So it was just like a, you know, US-grade software engineering experience, which I don't think, I think was very formative. And then right around the time I was 14, I did two things that sort of got me in trouble. One is I, I ended up reverse engineering Siri to make it work in Portuguese and Nuance and Apple were not very happy with me. And then the second was I built this app for the iPad called Quasar, which you could run, you know, apps on Windows the same way you can now on iPad. But that was in 2012, so 13 years ago. And that was really fun. And I put this online and I used to charge $10 per download. In Insidious Store back then, which was a sort of parallel app store. And I made like $300K when I was 14. So I had to explain to my mom what that money was about. She thought I was doing something illegal online. But anyway, so, so I did all that and then—

Can I pause you?

Yeah, go for it.

Because I read about some of this that I just had. I had a couple, couple questions along these lines. I mean, your mom goes to your first job interview with you. You said, I think you were 12 for that one. Like, what does that look like? You both walk into the office together.

Yeah. So, so it's funny because I think your parents make a few decisions that they may not be aware at the moment, but they drastically change the course of your life. And I remember around that time there were all these articles that came out that Bill Gates's daughter could only spend 45 minutes a day on the computer. And then my mom looked at that and here was her son spending 14 hours a day on the computer. And, well, Bill Gates knows a thing or two about computers, so Who am I to allow that? And I think she was very fortunate to see what I was doing with a little bit of a bigger aperture and see that I wasn't like playing games. I was actually building something productive that people got value from. And that sense of being a builder, which is really the biggest luck of my life, realizing that I love to build things, was something that I think she saw the impact it had on me and on the people that use the things I built. And I think that unlocked her to allow me to not only spend a lot of time on the computer, which I needed, but also at the same time, just really open to these experiences of going in and working, even though it was very counterintuitive. And, you know, my family and, you know, sort of extended family friends were all asking her, you know, this kid is 12, why is he working? And she's like, you don't understand, he wants to. And thankfully she did it because if she hadn't, I think the course of my life would have been very different.

What's— but what's that like? You go in for this job interview together and so what, the two of you are just sitting on one side of the table? So it was—

so we went to— we went to have lunch with the two founders of that company and then it was grandma.

Brazilian or Brazilian?

One of them is American, the other is Brazilian. Okay. The American one went to Brazil. Ben, he went to Brazil for Carnival one year and never left. I stayed there 4 years. Now he's in the US. He's in New York now. But it was really funny just to see, you know, like a professional software engineer in Brazil, in Rio. And then we went for this lunch and I was just talking to them technically about the things I was building. And my mom was like, I have no idea what's happening, but sounds like they are excited. So I guess it's a good sign. And then at the end, they were like, look, we think he's good. We'd like to work with him. And she said, I have only one thing to say. He's 12 years old. So if something happens to him, I'm going to kill you. And then, and then they were like, okay, understood. So she was very protective. But in a way, you know, obviously it was a very different environment. It was this company where, you know, I guess the second youngest person was probably like 26 and then I was 12. So I don't know if I would have the courage to let my, my son do that, but she did, thankfully. And, uh, and things worked out.

Uh, yeah. Okay. That's, that's kind of like how I pictured it in, in my head a little bit. I mean, look, lots of, lots of young kids code. Um, you know, and this is a, this is a difficult question cuz I'm asking you to, to tell me how awesome you are in some ways, but I, I, I'm more curious how your brain works because I, again, yeah, I've met so many people. They, this is a common story. They start coding at a young age. But I mean, what you're doing, it is, it is, it's again, it reminds me of HOTS just because it's on the exceptional side. How does like, what do you, what are you particularly good at when it comes to coding? Why were you standing out at that age?

I think, I think it was just this, you know, Paul Graham has his essay about, you know, the bus ticket theory of genius about these uninteresting obsessions that when you multiply over the course of one's life, you, you end up in a, in a, you know, in a pretty interesting place. And to me, I just had this deep curiosity to get computers to do things that they were not designed or supposed to do and almost sort of proving that it could happen from a technical standpoint. So that was probably the backdrop. And I would say the biggest benefit that I had having started that early is because I feel that when you are 12 or 14 or you're very young, the creative energy that flows through you when you're building something is almost like the purest kind because there is no other reason besides love for the game. Right. There's no other reason. Like, I didn't need to make money. Thankfully, my, you know, I grew up in a middle-class family in Brazil. So, you know, I was lucky to not have to, you know, I wasn't wealthy growing up, but it wasn't for the money that I was making. It was because I was just very deeply interested in the craft itself. So that to me has always been the thing and having this sort of respect for the creative energy that comes when you have an idea and saying, how do I manifest this thing into happening? And I remember when I built this app Quasar and I was 14 and it was actually a very tech, very complex problem from a technical standpoint. You had to make the iPad become, basically instead of being single processor rendering the UI, you had to make multiple processes render the UI. So you needed to build a scheduler to coordinate different things rendering at the same time on the screen, which was technically complex. And that to me was, I just had a sense that it was possible. And when you have this sort of deep connection to, this thing you want to see in the world, I think the way you show up and the hours and the time and the energy, these things all go away and you sort of surrender yourself to that process in a weird way. And I think so much of my journey as a builder was how do you connect to that energy? And it's funny because I was telling my team that, you know, now with AI and a lot of the things that we're building and being much more hands-on as an engineer, in a world of, you know, a company of like 1,300 people and a much bigger scale, weirdly enough connects me back to that very same energy that I started with and coming from a very pure standpoint. So I think that's been the biggest driver for me. And of course, you know, a lot of curiosity and energy and time, but it came from a very sort of creative place versus from like a work in the sort of traditional sense of I need to do this and make money and there's parts that are good, parts that are not good. No, it was all great. It was all really fun.

And so you're in Rio. I mean, all your peers are running around playing soccer, having a good time going to the beach and you're like locked in your house, like in a—

Yeah.

Yeah. And like in your room or—

Exactly. Yeah.

What did it look like? I mean, what kind of posters did you have all over the wall?

I mean, I, I had a— the time I made enough money, I bought myself an iMac. So it was me and my Mac and just like hours and hours and hours of like coding on, you know, Vim or TextMate back then was a text editor we used to use. And a lot of, lots of IRC channels. That's how I met GioHot originally. And, you know, a lot of the folks involved in Cydia and the original GeoBreak community, Mobile Substrate. There were a lot of different sort of work groups back then, but it was, it was, that was my universe. It was sort of living this parallel reality.

And then you were, you were kind of, you were a nerd.

Oh, yeah. Yeah, for sure. For sure.

If I, if I can ask, you know, when I was a teenager, both my parents got cancer and, you know, I'm sorry to hear. Thank you. And likewise to you. I mean, you're 8. Were you an only child?

I have a younger brother. A younger brother. 2 years younger.

Okay. And so, and your dad gets cancer. Mm-hmm. When did, I mean, if you're okay talking about it.

Yeah, of course.

Like, was this a multi-year protracted thing or?

So he had, he passed away when I was 8. He had cancer for like 6 years. So most of my childhood. Okay. He had a lymphoma, which is obviously a relatively rare kind of cancer. And, you know, cycles of ups and downs over the years. So moments that he was very well, moments that he wasn't so well. But it was a big part of my childhood.

Like, how do you— I remember, you know, in the moment for me as a teenager, I mean, I don't think it was obvious to me sort of how it was changing me. But, you know, a year or two go by and you can kind of feel it. Was it obvious to you? You're so young, or what is that?

It's funny because I I think I fully processed that maybe 5 years ago. Yeah. When you, when you sort of with the benefit of hindsight and perspective and living your own life and understanding that, you know, life is finite and all of that. And then you sort of appreciate the— you sort of resignify an experience you've had as a kid. And that was probably when I really realized. But for sure, in the moment, there were things that were that affected me, right? I think for a lot of this, this thing of building and coding, I think was how do you insert yourself in a place where you control all the variables and the entire thing is in your control because the outside world isn't. So I think there was a, there's definitely a relationship there. And look, I think my mom did an amazing job raising me and my brother after my dad passing away. But you see, you get exposed to, you know, yes, life is not infinite. You know, shit happens, that kind of thing early on. And that certainly shapes your brain chemistry in one way or another.

And so, and then your mom's a single mom from that point on.

Yeah.

And she was, I read, I mean, it sounded like your family was kind of doctors and scientists.

No, my mom was a psychologist. Okay. But she worked in, you know, psychology and then sort of marketing and, you know, more sort of a corporate life. And, and, you know, worked in a few different companies in Brazil as I was growing up. So, yeah, that was the backdrop.

And were you like pining for Silicon Valley or you thought—

Oh, no. So my original plan was to just like live in Rio, like most people that are born in Rio. And by the way, I understand that because the city is amazing. They never leave the city. So my brother, for example, he's 27 now and he's the sort of archetypical carioca, like the person born in Rio. And he would never— I think he's never going to leave.

Yeah.

And so that was my sort of destiny. And then when we started our first company in Brazil in 2013—

How old are you now?

Now I'm 29.

29. Okay. Okay.

Yeah. So when we started that back in 2013, That was the first time that I actually left Rio and moved to São Paulo and then decided to apply for schools in the US because, you know, I always wanted to build something at a sort of US-level scale. There were not that many big tech companies in Brazil back at that time. And we came in 2016 for, after building PagArmy, our first company that was effectively a version of Stripe in Brazil. And then we moved to the US in 2016 to start Brax.

And had you seen what the Collison brothers had done with Stripe?

So it's so funny. We, so we thought we were the original inventors of this idea of like, gosh, like payments are really bad for developers and engineers. Let's build a better version. And then one day we met someone and they said, oh, it's just like Stripe. And we're like, what is Stripe? So we met Patrick and John back then actually. We know them for a while.

And because they went down to Argentina or something.

Yeah, they actually, I think they worked there in the very early days of Stripe.

Yeah, but you didn't meet them on that show.

I didn't. I didn't. I think that was in 2011. I was still maybe doing too much jailbreaking, but, but, you know, it was— but we actually met them in probably 2014 or '15. That's when I met— I think I met Patrick for the first time. And we've stayed in touch over the years. And, you know, we have huge respect for them.

So you're doing this jailbreaking and then Pagarme is a payments system. I mean, did you feel this pull towards finance or it was just this problem that you—

So weirdly enough, the way I got into it is when I was 14, I actually changed jobs to a company in Brazil that was building sort of a version of Square, like mobile payments in your phone. And I went to work there initially to fix an iOS security issue. So, and there I was inside a payments company initially through this sort of iOS security angle, but very soon realizing how bad payments were and how old and archaic those systems were. And then I met my co-founder Enrique at the end of 2012, and he was in the sort of, okay, I want to accept payments online. That is a really challenging thing to do. There isn't a better solution. And then I was like, well, I was inside a payments company seeing how bad it was. So it was a little bit, you know, 1 1 3 from the perspective of the idea. And then as we started it, payments became this thing that we went very deep and started to understand better. And one of the things that I think was very big for us since that very inception was building from the bottom of the stack up. So we realized that at the end of the day, payments companies were limited by these vendors that they needed to use to actually interact with the foundational payment rails. So Visa, MasterCard, like the equivalent of ACH and Fedwire.. And as an engineer, to me, I had this like very almost religious view that you can never build a fundamentally better customer experience unless you control the entire stack all the way to the metal. A little bit the version of, you know, Steve Jobs used to say that, you know, if you care about your hardware, about your software, you should build your own hardware and just control the whole thing vertically. And we did that first in our previous company in Brazil, and we built the entire processing stack and the acquiring rails and Visa, MasterCard connections and all of that. And then we did the same again at Brex. And I think that was a big reason of a lot of the success and where we really started to innovate and being able to go after a lot of very large enterprise companies, AI companies, because of the vertical integration we built into the whole product. So weirdly enough, I think the engineering mindset shaped the way we ended up building both companies actually.

Okay. I'm gonna be gratuitous here and just lean in. Um, just Brex. Brex is our sponsor. You told me before we started you haven't seen our videos, um, that we make. And, and so I just wanna show you, I'm gonna try to do this artfully so we can have the camera.

Great.

See it. So you have not seen this before. You've not seen sexy brexie.

No, but this is cool. All right.

So we're doing an episode in Detroit and GM gave us a loaner.

Wow.

Car.

I want that. How do I get that?

So GM gave us this car for the week. Their lawyers did not properly, um, the, I, I found some, some wiggle room in the contract to do with the car as, as I saw fit. Wow. So we gave it to these guys to wrap it. Yeah. You've never seen this, but here we go. Meet Sexy Brexy. The Intelligent Finance Vehicle. Fan-fucking-tastic. Well, this subtle product placement is awesome. The truck is not really the point of this video.

That is so cool.

We are here to check in on the trade. Oh, I love it. I love it. You know, so usually around this time we have an ad read. Um, and I thought since you're here, I don't know if it's putting you on the spot. Would you, would you like to? Yeah.

What would you like to say about? I can read, I guess.

They only give me 30 seconds.

No, I'll give you 30 seconds. No, look, I think the reality is we believe that managing spend is a fundamental necessity of building a great company because you become what you spend on, right? A company, you literally are a byproduct of your spending and resource allocation decisions. And the fundamental problem is as a company gets bigger, You look at the finance team and they're responsible for 100% of the spend, but they only make 5% of the decisions. 95% of the decisions of spending in a company are made by employees and leaders everywhere in different offices, different countries. And that is an incredibly complex problem to coordinate. And a lot of what we believe at Brex is that typically as a company grows because of this amount of complexity is you have to choose either to put in a lot of controls in place and slow down the company because you need to have control to make sure the money's going to the right place, or you take off controls and then you increase speed. So speed and controls are this trade-off. And the reason we built Brax is to break that trade-off and give companies the agility of a startup all the way into becoming, you know, some of the largest companies in the world. That run on Brex today. So, you know, today we serve, you know, more than 35,000. 35,000 startups to the world's largest corporations. And, you know, 300+ public companies including, you know, like Zoom, you know, Palantir, Anthropic, OpenAI.

As of 2 weeks ago, OpenAI.

Yes.

And I've done this before.

Yeah. Robinhood. You know, you can go to our website and see the customer list. Which is really exciting. But one of the things we're really proud of is, you know, when we look into AI, we spend a lot of time making sure that we think about the role and how the finance function will fundamentally change. And all the big AI labs today and, you know, like Cursor, Vercel, of course, OpenAI and Anthropic, you know, and you go down the list and Mercore, You know, all these companies run on Brex because of looking at our AI vision and believing that this is the way finance is going to work. And of course, they looked at all competitors and all the alternatives in the space and still ended up going with us, which is a very cool proof point, I guess, of the way we're building the product towards the future of finance.

All right. That was longer than 30 seconds, but we'll allow it. We'll allow it. What did you think of sexy Brexi?

I loved it. It's really pretty cool, right? Very cool. I want to get a car.

We can make that happen.

That's great.

Make that happen. Okay. So, you know, you're— you meet Enrique. You guys apparently get in a fight, a Twitter spat at first, but then become best of friends and business partners. You both head off to Stanford.

Yep.

Together. It's kind of amazing that you both got into Stanford.

Yes.

Yeah.

He convinced me to apply, actually. I decided I I heard what an SAT was for the first time, I think was in mid-September 2023, and deadlines were in December. So that was a busy 2 and a half months.

Okay. Yeah, I bet. And then, and then so you, you get into Stanford, you guys, you don't stay around too long, about 8 months, I think.

Yeah. That wasn't the original plan.

Yeah.

But yeah.

So you did like a semester and a bit?

Uh, 3 quarters.

Yeah.

3 quarters. Stanford is a quarter system.

And so you're, you're already on to, um, you're, you're thinking about your next idea. I mean, the one part that I picked up on was that you guys get into YC, you start with a VR company. Yeah. And then it kind of pivots into Brex. I mean, that's a pretty dramatic shift. I was, I was curious about a couple things. I mean, as you're going through this very unusual childhood and working on these different projects, did you, what did you picture in your head? What was your dream like?

Yeah.

Was finance the end?

So it's funny because we moved to the US and we spent 4 or 5 years building fintech in Brazil, you know, 1 or 2 before Pagarmi, then 3.5 at Pagarmi. And the funny thing is, when we came to the US, we had this idea of, you know, fintech is a solved problem in the US. Brazil is this, you know, shitty country where, you know, fintech is still unsolved. That's why it was an opportunity. But here we want to be at the forefront of technology. So we went to this Microsoft store at Stanford Shopping Mall back in 2016. And we tried the first Oculus prototype, like an early, early version of Oculus. I think Oculus Rift or something, and we were amazed. So we thought, let's just go build that. And in a very naive way, we thought we could do it. And then we spent, I don't know, 2 or 3 weeks playing with the hardware and the software and technology. And we realized that, you know, we knew nothing about hardware, nothing about optics. And then we decided to pivot to something that we knew a little more, which was software. So it took us maybe 3 or 4 weeks to pivot again. And then in YC, we saw that all companies needed a corporate card and they couldn't get one. So they went to the banks and they got rejected. The bank said, you have no revenue history. And they said, yes, that's the point. That's why I'm a startup.

And even though you might have like $7 million in your pocket.

You couldn't get a corporate card just because you don't have this track record. Exactly. And we thought, well, but they have cash in the bank, so what's the risk? And the banks didn't get that. So that was the original idea that allowed Brex to scale, was underwriting the companies based on cash and effectively build a better American Express. That was the original, original thesis.

And so it's supposed to be Brex, Brazilian American Express. Instead of Amex.

Brazilian Express. Yeah, that's— so the myth goes. But yes, it was a 4-letter domain pronounceable in English with a backdrop name.

Okay.

Backdrop story for sure. So that's how we got it off the ground. And it turns out it was one of the things that was interesting between Brazil and the US is there's actually this weird negative correlation between how developed an economy is and how I think bad their payment rails are. So when I was born in Brazil in '96, we had real-time settlements. Like the day I was born, it was not cheap, but you could settle a wire in real time. We're in the US now in 2026, right? 29 years later, and you still don't have, you have like partly, but you cannot, like, not all payments are in real time yet. And the reason is because when you look at Brazil in the '90s, we had hyperinflation. So the inflation was 30% a month.

So you had to clear it.

So if you didn't clear in 2 days, you lost like, I don't know, like 5, 7% of the value of the money. So there was a very deep investment in the payment rails because the economy needed it. And then the US, because this is such a stable economy and such a stable currency for, you know, hundreds of years, I don't think was as developed. So weirdly enough, we saw a lot of room to, to actually revamp a lot of the financial stack in the US in a way that was bigger even than Brazil, just because of the scale of the US economy.

People explain this to me sometimes. I mean, you know, if you go to Europe, they're doing all kinds of amazing, like very interesting stuff with financial systems. When I go to China, it's insane. Yeah. Even like I went to Nigeria to shoot an episode of our show. You know, there's parts of it that are like really backward, but then there's other parts where you're, you're just paying for your cell phone at like whatever, the chicken shop. And, and, and then, you know, South America in Mexico are quite famous for innovating on financial tech. And so the US always seems to be behind. It reminds me of like when I was starting out as a reporter, we were just woeful at telecommunications and Europe was so far ahead. And then, and then like the iPhone like sort of rescued the US and made it interesting again. But I mean, people have told— you went over some of the points of this, like stability. But other people said to me, it's because we have our banking system.

Yeah, we have thousands of banks in the US. So for example, like Brazil, the central bank created this payment rails called PIX, which is effectively a real-time settlement between any two accounts. You can use your phone number or your Tax ID as a key, and then it finds your bank account. I can transfer money to you right now, like almost any amounts, which by the way, has other security implications in Brazil, which the system ended up handling later, but that's a separate story. And then in the US, but in Brazil, you have really probably 30 major banks. In the US, you have thousands, like 2,000 or 3,000 banks today. So when you want to roll out anything new at a systemic level, you have to go in bank by bank and make sure they comply. So even if you look at RTP, like real-time payments today, the ubiquity is, of course, the big banks support it. But if you're transferring to a community bank or a regional bank, a lot of times they don't support because technology is really behind in terms of adoption. So being more concentrated also allowed for a faster innovation cycle. So, but I would say it was the fact that the economy really needed it, right? In Brazil, we, because a lot of the payments were in cash and offline and even on cards, the government saw this need for first sort of bringing and modernizing payments. And the central bank in Brazil, I think, did an amazing job with that. And then second was moving more of the economy outside of cards into something that the government had more agency over., and, and these movements were easier just because of like, one, the pain point that existed, and second, the concentration of banks.

Like, I'm going like, I was gonna go chronologically, but now I just wanna talk about this for a second. I mean, so the US is, is like destined to be woefully behind the rest of the world on, on many interesting financial innovations. Like we're, we're stuck unless we, we kind of change this.

And the US has this like, has a lot of, sort of particular things that, you know, when you look at the amount of the economy and credit cards, for example, of course you have, you know, you have rewards, which is a very big part of the decision-making factor. You have, you know, no real-time payments in some things. And especially most money movement is actually not real-time today in the US. You still have a lot of ACH. That takes 2 or 3 days to clear. You have, you know, there's no, for example, something you have in Brazil, which I think is fascinating, is you have a clearinghouse for receivables. So because the interest rate's so high and the country is so dependent on factoring of any receivable you have when you have a contract or something, you actually have a centralized clearinghouse that you can register a contract and get a prepayment on that contract. Doesn't exist in the US. So there's a lot of things that are, that are very different. But at the same time, it creates a lot of opportunities for companies like us to, to go in and do new things.

And I— okay. And then, you know, this next question comes with some baggage because you've just— you're part of— you're going to be in whenever the deal closes, part of Capital One. It's a soon, you know, traditional financial player. I just always found this shocking, you know, because I followed Stripe. I knew Patrick and John when they were like, I don't know, there's like 8 people in the room or something like that. And, and, and then just the story story you were describing about startups had money in the bank, people wouldn't lend to them. I mean, like, it's sort of— I mean, in some ways it makes sense because this is the story of business, is that incumbents move slow and don't see opportunities. In other ways, I find it really shocking.

Yeah.

That like Stripe could become such this foundational layer, this massive multi, multi, multi-billion dollar business that, that the traditional financial companies were just apparently, you know, willing to, to see to these two Irish kids. You know, Brax just took off like a rocket ship from this, this insight that you guys had and you being able to code all this infrastructure to, to make it happen. Yeah. Like, like, are there still— are there— is that typical? This is the same as I would see if I went into the energy industry or defense, or is finance uniquely kind of conservative and bad?

I think, I think it's, you know, if the history of technology serves us any backdrop, I would say I don't think finance is too dissimilar. What I would say is I think the interesting aspect is first, just the scale that finance has in terms of how businesses operate and how how ubiquitous it is as a part of what makes just the US economy work. So for example, you look at Brex and we created this new category of modern software and financial services together. When you look into the US economy today, we made a lot of progress, but 97% of the US economy still runs on traditional corporate cards, 97%. And there's a lot of room to grow in terms of just expanding what we've done and serving the rest of the market. And when we, when we, when we, you know, got approached by Capital One, the interesting thing for us was we started to think, okay, where can the company be, let's say, in 5, 7, 10 years in a standalone path? And then what if we, we join forces with yes, an incumbent, but actually I think the most tech-forward bank in the world that, you know, really realized that you could bring in data and machine learning and technology to revolutionize credit cards and the way underwriting was done. And by the way, there's sort of fascinating stories of how Capital One started that and what could we do having that platform at that scale. And as we started to click into that, it was impossible to unsee it. And really the way to think about it is how do you bring the scale, brand, credit of major US financial institution with the technology and product we have? And in a very interesting way, the thesis behind this is exactly what you were describing is how do you accelerate the adoption curve of this technology that we believe should be core to how every business runs, but do it at a scale that is country affecting, right? That you actually inflect the outcomes of so many businesses that depend on these things to run and operate and are still working, you know, in, in with technology of like 30, 40 years ago.

Hello, geniuses. Let me tell you about E1 Ventures. They are a venture capital firm in Silicon Valley, a longtime supporter of this podcast and a longtime supporter of Big fantastic world-changing ideas. If you have such an idea, hit up E1 Ventures or send me a note and I'll put you in touch with E1 Ventures. It's, uh, it's quite a, quite a deal for listening to this podcast. Um, thank you again, as always, to E1 Ventures for their support. So like, and these things, I mean, they do work sometimes, these types of arrangements. It's also not uncommon for the big incumbent company to, to reject, you know, the, the, the, the innovation arm that they have purposely set out to acquire, you know, for that very— yeah, reason. Yeah.

So it's funny because one of the things that— well, two things I was very impressed by Capital One. One was just the speed. So from inception of this idea all the way to signed definitive agreement was a little over 40 days. So they actually moved with a lot of conviction and speed. And second was the fact that being a founder-led company really changes the calculus for how you do things. And speaking to Rich and the Capital One team, one of the major things that they realized was, look, the most important thing for us to think about as we make this work is don't crush the butterfly. So how do you leverage the scale and the distribution and all the benefits of Capital One, but also keep the things that are uniquely Brex and made Brex into what it is today alive and growing. And the way Rich, the founder, describes it is, how do we just add water to the Brex plant? And of course, be very mindful about the risk management aspects and the aspects that are inherent to being a bank, but also leverage the scale and the distribution. So, I would say there's been so much thoughts on how to make this as lightweight of an integration as possible to preserve the momentum that Brax had as a standalone company and actually accelerate it. So, you know, when you look into the outcome here, we really think in 3 to 5 years, the company will be just dramatically bigger than what it would be on a standalone basis because of the the water and the scale that the capital one brings.

Let me ask one annoying reporter question. I mean, you at one point, you guys are valued at $12.3 billion in 2022. It's a different world for a variety of reasons. You sell the company for $5.15 billion. I mean, look, I think anyone would be happy, happy with that outcome. But it is the nature of the beast then that some people compare your previous valuation to this number and we're in this, this like total frenzy of, of you guys, some companies that shouldn't be named, you know, all competing against each other very, very vocally, very aggressively. I saw some, some RAMP cheerleaders, you know, celebrating this $5.15 billion number. Yeah. I mean, it's got to be a weird feeling to like have— you're having this great moment.

Yeah.

There's a lot of money, you've accomplished something, it marks a moment in time. And then, you know, I guess it's competition, but then for it to be portrayed by some people one way.

So I think there's two sides of this. So the first one is one of the things that we realized in 2023 is the company was in a very different position than today. And we decided to sort of aggressively turn things around. And that was end of '23, early '24. And one of the things that we did is that was very unpopular and externally back then is we actually did a big reset on the internal valuation. So, you know, yes, of course we raised $12 billion. That's, you know, that's a fact. But at the same time, how do you create enough upside for employees so that a $5 billion outcome is a win?. And that's what we did. So we went to the board, we actually repriced all the employee equity at a much different price. And that effectively really changed the picture for employees in the team. And, you know, investors that invested at $12 billion, of course, you know, got their preference back. So it ended up being actually a very good outcome for everyone. But the key thing was in the moment where things were the hardest, like 3 years ago, actually looking ourselves in the mirror and saying, look, you know, I always say, you know, hard decisions, easy life, easy decisions, hard life. And the hard decision was we let go of, you know, almost 30% of the company. We repriced the equity, we changed the way we operate a lot. We promoted a lot from within. We took out two layers of management in the entire company and we started this thing that I called Brax 3.0. So that backdrop was how we got to where we are today. And by the way, where we're still accelerating even faster than when we closed the deal, when we signed the deal back in January. So I would say the best way to think about it is we made a lot of deliberate decisions 2.5, 3 years ago that made this a win for employees because we decided to face reality and and actually look at it with the lens of everything eventually converges to public markets, whether you like it or not. That's the only price signal that really matters. So how would you think of the company today if it was a public company? And how do you ground ourselves in that reality? As hard as and painful as it might be, because the fairy tales of 2021 were just that, fairy tales. So then, and then I think the second piece is, of course there's competition in the space. It's a massive part of the US economy, massive part of fintech. And it would be ludicrous to think that there wouldn't be. But I think the best way to think about it is, you know, us and all modern competitors in the space, we're still 3% of the US market. So the question becomes, how do you go after the 97%? And what we decided to do at Brex is we said, look, the reality is the chances of hitting two orders of magnitude bigger scale together partnering with a massive incumbent are fundamentally different. And as we started to actually just see the scale of like, for example, like Capital One spends $6 billion a year in marketing. We spend around 1% of that. $6 billion in R&D, we spend 2 or 3% of that. So when you just apply the scale into the problem and it would just accelerate what's working, The question for us was how do you solve for the 97% and how to be the clear winner there and subordinate all decisions to that, including structure, the capital, you know, all of that. Because for me, it was always about at the end of the day, I wanted Brex to become a major part of the US economy, Brex to outlive and outlast me. And the same way I look back, I left my previous company in Brazil 10 years ago and Pagarmi is still second or third largest player in Brazil in online payments. I want the same to be true of Brex and I want to be here for the next many years. But still, I think that the scale, solving for scale and just the reach of this technology in the way the average business runs was fundamentally a different thing to solve and it's still very early days.

Well, please introduce Capital One's marketing budget to Sexy Brexie and let them know what's possible. Yeah. And okay. And, and as far as like you've been on this, I mean, it just feels like a whirlwind, man. I mean, you're starting, you know, a business around 12 and then here you are, you're 29. I mean, it's been nonstop one thing after another. You've, you've been actually like, to your credit, super open. You write these essays from time to time about, about big moments in your life, mental health. Things, um, things about— you had this really nice note about your mom. Um, and I know you talked about some of the decisions she had made, but you know, she seemed, um, like a kind of like unconventional thinker around, around a variety of things with this, this kid who was, who was living an unconventional life. And so anyway, I found you're like really, you're quite thoughtful and open about these things. Um, Do you— is there like part of you though? Yeah. That just feels like you've been on this, you've just been carried, you know, not like you were, um, not an active participant, but I mean more that you've been bounding from one thing to the next on this ride. Like, is, is there— I always ask the Collisons this. Okay. Cause they're like two of the smartest people I know. They are, um, polymaths. They're interested in so much different stuff.

Mm-hmm.

I know financial tech is very important. It's not the particular thing that like, yeah, I think about all the time. And so, you know, I'm always like, what do you guys, what do you guys like want out of life? You know, because, because is this, is this just the thing that you made when you were young and it turned into this huge hit and you're, you're trying to, they always talk about increasing the GDP of the internet and the world. And I, I see that, right? It opens. I went with Patrick to Israel and Palestine and I saw people who were, setting up businesses on Atlas that, yeah, like otherwise just could not incorporate their company and now they're selling to the entire world. So like, I get it. I'm not trying to undermine something, but yeah, like, is there part of you that just feels like you've— are you making all these conscious decisions or you've just been pulled on this journey?

It's a great question. I think, I think, I think it's a little bit of both because I would say On one hand, you know, I really like to believe that, especially now, you can, with AI, learning new things is a fundamentally lower barrier than it's ever been. So if I said, look, I want to learn how to make X, I think it's possible to learn and master that. At the same time, I think the sets of opportunities that exist that are interesting they tend to be related to domains that you explored in the past. And I think your ability to recognize what's a good opportunity is also higher in things that you have more exposure to. So I don't think we would've been able to build Brax if we hadn't built our first company in Brazil because we just knew too much about payments. We were just very deep in the weeds, we knew the issues with credit card issuing. We knew what the acquiring side looked like on the merchant side. We knew a little bit of credit underwriting. Our first company was also regulated by the central bank and we were the officers of that. So there were a lot of conditions that made Brex possible. So in a weird way, I think it creates the set of experiences that make the next thing possible. And I think there's something that Jensen says that I think is really true, which is You know, at the end of the day, it's easier to learn to fall in love with what you do than to do something that you love. Because I think when you think about love and sort of a deep connection to a craft, ultimately something that I think is very deeply encouraging is when you feel the sense of mastery with something and not from an ego perspective, but from the perspective of I just understand every layer of this problem and I can think about second order effects. I can think about implications and I can see the world through a much bigger aperture because I have this unusually high exposure to this problem. And, and I think there's something about that in fintech where, you know, the more you think about, you know, and we see this with AI, like we, you know, we spend a lot of time looking into what actually changes when you start to deploy AI agents in companies. And of course, the early things are, yes, there's a bunch of cost savings. Second, there's compliance increases and data quality increases and automation increases. But then you look into a layer deeper and you see, okay, how much time does the US economy spend managing expenses? And turns out it's 50 million hours. And then we're like, okay, how many things can we do possibly that contribute 50 million hours to the US economy? And we'll build these agents that automate all of the T&E process and you see the impact inside a customer and it's pretty remarkable, the reactions, the finance teams where they say, wow, I actually didn't even think this was possible even with technology today. And this is like, by the way, on Anthropic and OpenAI. That are sort of at the bleeding edge of AI and deploying AI internally. So, but then it's like, okay, how would we even recognize that that opportunity existed? And it goes all the way down from, well, yes, we were building corporate cards originally, and corporate cards are a big part of how spend gets initiated in the company. And it turns out that initiating and making spend decisions requires a lot of sort of delegated decision-making. So you need to create tools that empower the financing to be in every decision. What can help with that? AI. And then, well, how to make AI really effective at scale and give the financing control over how AI is operating. And then you put a lot of agents. So you sort of cascade this down and there's 7 or 8 hops from the original idea all the way to the outcome that we deliver today. And these things were recognizable because of a previous pattern. So I think that's the part that's hard to overstate. I wouldn't say that it's— I'm sort of a— it would be irresponsible to not do something that's not fintech. But there's something to be noted about the fact that you get to certain places because of where you've been in the past. And I'm grateful for the experiences that sort of exposed me to the the issues and the problems and the solutions that I was able to build by having been on this track.

And do you think this is going to— maybe it's really hard. I mean, you're still so young, 29. Yeah, I read about you being obsessed with space documentaries when you were a kid.

Yeah.

You know, I hit you up on a, on an AMA asking what you were interested in outside of fintech, and you mentioned space and energy. Do you think there's like another chapter in your life where you dig in on something like that, or is that something maybe you're investing in and, and yeah, it's on the side?

I mean, I am in the, in the sort of a spirit of both topics. You know, I'm actually very curious about robotics now and trying to understand what's at the frontier. And of course you see Tesla and Optimus and all that, but also what are sort of more bootstrapped versions of that and new companies building things that are pretty interesting. So that's been a big area.

Meaning like you just keep an eye on it or you're—

I'm getting mostly keep an eye on it and try to angel invest in a few things. But I would say that's sort of a newer curiosity over the past maybe like 3 to 6 months.

Okay.

When you look into like things I'm spending a lot of time is, is, you know, I am really interested in this idea of virtual employees. So how do you actually deploy virtual employees inside a company that look and feel and act like an employee? So they are real people with real names on Slack, on email, engaging with people, joining meetings. I got obsessed with Open Claw over the past like 3 months and, you know, automated my entire life. I can tell you more about that. But these are the things that I'm spending a lot of hours now.

Oh, no, please do. Do you have a human admin?

I do. I do.

And then is this human admin competing against your virtual admin?

They— I would say they don't overlap as much, deliberately. My human admin is amazing. But I would say no. So look, I think the how we got here, right? And I think Opus 4.5 was AGI or very close to AGI. And Opus 4.6 was the sort of missing, the sort of rough edges that got a little better. So the way I think about it is electricity was invented in December. We're now in March. What are the things that you can build with electricity now? And that is the big question. And to me, the fundamental capability jump was that of course coding models were the very interesting thing that, and you see Claude code and all these agent harnesses evolving. But I actually think the most interesting property that was unlocked was the capability of a model to self-bootstrap. So basically one of the things that we're experimenting in IBRAX is actually using Open Claude, to create virtual employees. So we created this like virtual recruiter, for example, we call him Jim. And Jim is literally a Slack entity, a Slack person, has his own email and engages and talks to recruiters internally. And the thing, the whole challenge was how do you build Jim without writing a single line of code? And the idea is, and of course there was still semi-technical people involved because it was one of our first early iterations of this. But the really interesting thing is you see a recruiter engaging with Jim and then saying, you know, for example, I just saw this before coming in here. A recruiter was saying, look, can you take a look at this person's application? I have a sense that this resume was fabricated. So how do you look at that? And then the agent looked at that and actually built a capability of screening a resume and understanding if it was fabricated or not. And that was never coded by anyone. And I think that's the fundamental leap that we were missing. And a lot of what we're spending time thinking is how to deploy this safely, of course, inside Brex. And there's a lot of things on the security side that we're also spending a lot of time because I feel the challenge that the industry is in now is people say, well, you can't predict what models do, they're not safe, therefore I'm not going to use something like OpenClaw in a very important environment. Because who knows what's going to happen. And the stance that we decided to take is saying, well, why don't we just solve for that security aspect and then actually deploy these things internally and leverage that technology and see where the frontier really is. So we're actually going to open source this, but we actually built this entire technology layer. We actually call it Crab Trap, funnily enough, where the idea is when you have something like Open Claw running, we, the best way to monitor what an agent is doing isn't to have a human looking at it, is to actually have another LLM looking at it. So we build this thing that effectively intercepts all the traffic coming out of an Open Claw instance and routes it through another LLM that's through an HTTP proxy that screens all the traffic and says, is this something that a recruiter agent should be doing? So for example, should a recruiting agent send like a harmful message to a candidate? No. And it can go in and block that request at the network layer while the agent itself is not even aware that that's happening. And the really cool thing of that is you can actually run this at scale. And the only technology that we think will be able to monitor agents is actually agents themselves. But you build this almost adversarial effect where you have one agent monitoring another agent. And we've seen surprisingly good outcomes of this thing so far. So does that mean that we're deploying, you know, Open Claw and virtual employees in the most critical workflows in the company? Not yet. We're deploying a few. But the really interesting thing is instead of saying, well, let's hold back and not deploy these things because at the frontier we haven't resolved security yet, we said, no, no, why don't we just solve security, which is a systems problem. It's a distributed systems problem more than anything else. It's an engineering challenge. And then the technology and the LLMs and the state-of-the-art of the models is already wildly capable, except for this one problem that we'd rather solve around. So that's been fascinating.

And then you were saying, and you're using OpenClaw in your life. I mean, is this the stuff you're talking about?

Well, no, we're using OpenClaw in a bunch of different contexts. Virtual employees is one. The other one is I use it personally to do everything.

Like what?

At this point, I literally run Brex through Open Claw right now. So the way it works is I think most— there's a lot of productivity tools of AI, right? And I think the thing they fail is that they don't assume that the world is a very noisy place. So when I go into Brax, Brax has like thousands of Slack channels, like thousands of people, thousands of— I receive like hundreds of emails a day. And the question is like, what matters? Like what to pay attention to? So I build this autopilot system that starts with this signal ingestion pipeline. And basically the way it works is It screens my email, my Slack, Google Docs, WhatsApp, a few different data sources. And then— but the really interesting thing is I decided to build my automation in my life around two concepts: people and programs. So programs is almost like a project that I want to be updated on. So for example, financial performance of the company, Capital One integration. AI strategy internally. And the thing about a program is it's very declarative about what I care about and what I don't care about. And then what it does is it screens those signals in the context of these programs and then organizes interactions that happen that I should be aware or action items. And it does the same for people. There are also people that I care about. So I declare who are the people that I care about and there's probably 25 people in the company. In my life more broadly, I declare those people and what are the things that I actually care about. And then the signal ingestion filters with the lens of these people, it denormalizes signals and then it processes them. And then it produces this very clean summary of everything that's happening in the company that I should care about and all the action items. And then there's a really cool part, which is all the action items that I should do, are things like, well, I finished the customer conversation, I need to go in and update this, the deal team on that account, or I had a meeting, someone asked me for an intro, or all these things that are follow-ups from meetings. And the cool thing is because I have the signal ingestion, I can go in and I have this other thing called an auto-resolver that it can go into the list of to-dos that I have and say, how do I automatically action this task based on the context from the meeting?. So for example, let's say I go in and I say, we should talk about, I wanna figure out how to get that car. Let me have a follow-up on that. So then I can have Granola listening to the conversation. Granola is running on every meeting that I have, ingesting that into the signal ingestion pipeline, creating an action item. And then when the auto resolver runs, it says, okay, what's the context for this to-do, this task that I have? It will go back to the notes and then it can say, well, I can send you a text or send you a Slack message or you an email. Depending on, you know, whatever is best. And then it can draft the message automatically. And then all I have to do is click a button. So all of these are building blocks and skills that you can build together. You can, you can do it either on OpenClaw or Claude. I actually have it running on OpenClaw now to effectively decompose the CEO job of like ingesting signals, what matters, building a summary, determining tasks, making those tasks happen, monitoring my team for those tasks. So it also can do follow-ups automatically. And I can also go then and build skills. So for example, I can go in and say, well, a way to auto-resolve is a Slack message, or is there's a skill that I created called Review a Doc as Pedro. So there are things when I'm reviewing a document or a product review or a presentation that I always ask. So for example, what is the most important thing that this thing should be driving and how did we choose that thing? What's the bottleneck making this thing happen to go faster? Why aren't we moving faster? There are like 3 or 5 questions that I always ask. So you basically make that into a skill and then you build that into the pipeline. So it's basically Lego blocks that you can put together to accomplish really complicated things through these systems. And of course, as an engineer, I had to go and build these things. And I think I'm going to open source this, but it's basically this entire CEO or I guess like almost like, you know, leader in an information-based company workflow. And it's been really fun to build it.

So it sounds like, I mean, you just got really sucked into this and you just mentioned, I mean, you're doing this yourself.

Yeah, like it's the AGI psychosis, I would say. We have a rule at home, which is no AI Saturdays. My wife is like, because like I'm basically like talking to, my Open Cloud all the time. I'm sending voice notes and it's the best developer experience because it can be on the way here. I was like, I had this synchronization issue on my data pipeline, so I was literally sending voice notes saying, hey, change this, change that, do this, do that. And it's amazing. I can be on the car. I was in a Waymo coming in here and getting a bunch of things done. So the problem is it's so easy that you start doing it all the time. So then I can be having dinner and sending voice notes and my wife's like, stop, let's have dinner. And then Saturdays, no AIs, no AI at home, which is, which is kind of a fun thing.

And when you were, when you were living such a frenetic life and, and, and getting pulled in all these different directions and you kind of, you know, you talk, I've had a panic attack. You talked about having a panic attack. You talked about getting into meditation.

Oh yeah.

And reading and sort of changing your nighttime ritual. Like, what kind of things do you read at night?

Gosh, right now I'm reading a totally different thing. I'm reading Benjamin Franklin's biography by Walter Isaacson, which is actually quite good.

I've read that one. Yeah.

Yeah. Very interesting. I mean, I generally like— The Da Vinci one is also very good because I think there's this thing where In a weird way, a hundred, a couple hundred years ago, or even longer, I think the sort of Renaissance man, this like multifaceted person that knew a little bit, or actually even a lot about different domains was a big thing. And intersection of arts and science is something that I like a lot. And, you know, Benjamin Franklin was a, not only of course had a massive effect on as a founding father and the definition of what the US is, but also as an inventor. And he was so deep in electricity and things like that. And that's very remarkable.

Journalist.

Journalist, of course. So, and ambassador for the US in France for the first time right after the the independence. So, so it's, it's remarkable to read it.

Is that like, do you like nonfiction?

I almost always read nonfiction. Yeah. I read your book, the Elon book, which is very good. And you can— and I think, I think so much of what I enjoy is trying to understand someone's frame of mind. And one of my favorite questions, it's a weird question, but it's what is the experience of being you? And it's a profound question because there's a few like mental models in your head when you look at the world that sort of shape your experience of acting and being a person. And I think that's very profound. So when you read these like biographies that go very deep, I think ultimately what you're getting to is like, what is the experience of being this person? And great biographers, as I'm sure you know, like, ultimately, that's what you're trying to crack, right? You're trying to— there's all these stories, and it's almost like, you know, there's this elephant that you're touching different parts and you're trying to triangulate what is the entirety of this person. And I think biographies are great at that. I mean, I read a few technical books on The Art of Doing Science and Engineering from Stripe Press. I read a lot of things that are relatively technical, but I think biographies and things that go very deep, The NVIDIA Way from Bob Jensen, I think gives you that same mental model. And ultimately, I think it's the Charlie Munger thing. At the end of the day, being successful is a collection of like 30 to 40 to 50 good mental models and just applying them. Everywhere. And I think seeing how remarkable people have different sets of these, like 30, 40, 50 different mental models, I think is pretty fascinating.

Yeah. No, biography.

Yes.

We could talk for a long time about that.

Oh, God.

Very strange experience doing one as well.

What was it like? What was the— what were the—

God, I set myself up for that, didn't I? You know, I mean, I really could talk for hours. It was, it was It was funny in some ways because I, at that moment in time, I wanted to do a book. I'd gone to see SpaceX. I was working on a magazine story about Elon. I had not like set out to do like a biography on Elon Musk, but I, you know, I, when I went to SpaceX, it was, it was just so, so different and further along than I pictured in my head. And it was just right in Los Angeles, and it was, it was like, it was thriving and just had this energy. And anyway, that's what kind of sucked me in. And then I got interested in Elon, you know, interacting with him and reporting the story. And then, so anyway, you know, I signed up for this thing, but I, I had no idea what I was getting into in the sense that you spend 3 years thinking about this one person, and that's like your job. And it's like, it's very all-consuming, you know. You start to dreams about that person. Um, well, all the things you mentioned, you're trying to do, hopefully you're trying to do a good job. And so you're not, you're trying to, you're trying to talk to a bunch of people to get at what you're talking about. It's like, it's like Elon's always going to represent sort of like how he wants to be represented. And then you're trying to talk to all these other people to triangulate. I don't know if you ever get to truth, but some, some version of that that you're conveying, at least you can you can give people, well, this is how these other people experience him. And, you know, and try to create as full of a picture as you can. But then you get into this, it becomes very strange, right? You, you, I, you know, it's like I'm thinking about him all day. You're like having dreams about him at night. I was like, I didn't sign up for this. I want some peace, you know? And so, I mean, there was just part of it that I didn't expect.

That's interesting.

A lot of what you're talking about though, you're, you're like trying to put yourself in this person's shoes. And then, then there's this strange thing where you're like, you're, I don't know. I don't know if I was ever really trying to pass judgment, but you are passing judgment on certain things for the people that are gonna read this book. And then it's, it's just like a lot of, I dunno if it's like pressure, but, but if I think if you're wired the right way, you kind of, you wanna do a good job at it. And, and so yeah, it's very strange. And then, um, after that, you know, that book became such a hit that there's just all these things I didn't sign up for, you know? Now you're like the Elon guy. And it's like, it's like, yeah, I found it fascinating, but I'm interested in lots of shit, you know? Yeah, yeah, yeah. I don't, I don't need to be the Elon guy for 10 years.

Yeah, 100%. And so, but, you know, I think the gift of, you know, great biographers is, well, just books in general, right? Is I don't think people realize how remarkable it is that you spend 3 years of your life dedicated to this, and I can sit down in 4 days and consume that. Yeah. I think that's such a profound, such a profound realization.

That's a crazy amount of work.

Yeah.

When the Heavens Went on Sale, the one I just did, that was 6 years. Oh God. It's just, it's like, it is, it's kind of crazy. I mean, I love it. That's actually the most fun part is the journey, you know, and like the process and It's amazing. By the time you finish the book, you've worked on it for so long, you've read it so many times, you're just like, I never need to look at this again. It's just like, now I want to do the next, next thing. But that, that exactly what you said is also what I love about magazine writing. It's why I liked working at Businessweek for so long was that I got to go in a, in a condensed, smaller form than the book to go like live your life for some version of your life for 3, 4 months. And and like inhabit these worlds, right? And like see what people's mental models are like and see what they're, um, you know, it's not always perfect because again, people are putting on personas and things like that. But if you, if you spend a decent amount of time at it, and I'm sure the more you do it, the better you get at cracking it a little bit, right? I think.

Yeah.

And, but that's the best part of the job is like, you just get to see, it's like, oh, this is how like Brian Johnson lives. This is how Jensen lives. Yeah. How, you know, uh, Jennifer Dowdle lives. It's like, yeah, it's just, it's Cool. It's awesome to sort of shed your own ego for a bit and just, just watch. Yeah. Yeah.

That's inspiring.

Well, I want to be mindful of your time. I just, I just have like a couple last, I mean, they're more just like just out there questions, I guess. I mean, I, I'm just so curious what you think like the future of money is in 20 years.

Yeah. I think, I mean, I think you're going to have a lot of the sort of payment evolution, right? You see Visa, MasterCard already positioning themselves on stables. You see, of course, Stripe doing a lot of moves there. So I think that's going to continue, but I actually think that's less interesting. What I think is more interesting is the decision to spend money itself. And I fundamentally believe that, you know, you become what you spend on as a company, as an individual, you become your resource allocation in terms of the people you hire, in terms of the decisions you make, the technology you use, right? You know, one of the quotes that I really like is, you know, we shape our tools and our tools shape us. So I think the way— I think there's going to be a fundamentally higher level of thinking behind how companies make financial decisions. And you can imagine that a lot of it is this problem of context and just aggregating information and understanding what are the implications of this decision to, of course, the revenue, the costs, but even to things like the culture of the company, to the vision and the mission and what are you trying to do. And these things cascade down all the way into a single individual Wednesday, right? So there's this thing that I forgot who said it, but life is a picture, but you live in a pixel. So at the end of the day, you can talk about these visions, but it has to materialize on a mundane Wednesday. Like how does that change your mundane Wednesday? And I think when you have this autopilot to the way companies make decisions, and you fundamentally change the calculus of where capital is going to be more aligned to where what matters to you as a company and as an individual, I think that's a very big unlock. And I think it's going to be more profound than people realize. And of course, AI is going to enable that. And you're going to need, of course, a lot of the payments technology, agentic payments like agentic commerce. All these things will play a role.. But ultimately, like, you know, what's the icing on the cake and what's the cake itself? And the cake itself to me is this idea that, you know, the quality and the ability and the level of discernment that you're going to have in financial decisions, because you can have so much of this context, it's like, what if you had the best CFO in the world making every decision? You know, the best, you know, founder in the world, helping you make hiring decisions. What if you had the best person in the world deciding where to, where to expand next, where to grow, how to grow, how to look at your unit economics? And I think that will become true in the next few years and hopefully we play a big role in it, but it, it's like it's gonna happen regardless.

So then companies just like, if everybody has some sort of perfect information and they have the ideal CFO and the ideal, all these positions, I mean, so you just still end up competing on, on, yeah. Your product and your—

So that's the fascinating thing, which we have had a lot. So for example, we were building these recruiting agents and I think the models are, because of the amount of post-training that goes into them, they have this relatively uniform view of what is a good candidate. But then the question becomes, what is a good candidate for us? And I think it's like focus and constraints are what give meaning to your choices. So when you have to go to a recruiter and you say, what are you willing— like, where does this person need to spike on and how do you hire for that instead of sort of lack of weaknesses, which is where the models generalize and go typically, right? It's very hard to make a model say no. And I think that's a very profound learning because that is the hard question. It's like, where does like spikiness matters? Right? Where does it matter to be astronomically different? And I think that's where the judgment comes in of what kind of company you want to build. And I think that's an inherently human decision.

And you're giving it that flavor.

Yeah, that's the challenge. It's like, how do you give that taste of Brexiness?

Yeah, because there's like a world where the AI recruiter doesn't want you or George Hotz because you look pretty weird.

Exactly.

Yeah.

And one of the things that we do is like one of the factors that we really like is I mean, of course I'm biased by this, but it's who were people that did something at a moment in their life where they had absolutely no other reason besides love for the game. Yeah, love for the craft. And well, people that are coding in high school don't have any reason to code, maybe getting into college, but you can just tell based on the kinds of things they build and the relationship they have with the craft. People that were doing completely different jobs, like someone that was in finance but ended up going super deep in AI and cloud code and built this massive thing to automate their jobs. They didn't have to do that, but they did it anyway. That was driven by something pure. So that is something that we care about a lot and we want to teach our recruiters to do it. But that's us. That may not be true for every company. And I think this ability to understand the ethos of what you care about when it comes to this function and separating that from the technical excellence, I think is a really important thing because the technical excellence I think will be true no matter what. And the point about the ethos, the reason a model can discern that is because that is a choice and you can't really argue with, is it better to hire someone that started coding before they had to in their lives or to hire someone that went to academic route and has a PhD in machine learning and AI and is going to go do research at OpenAI or Anthropic. It depends, depends on what kind of company you want to build. And I think that's the part that's really interesting is if you just take out all the things that are not about the ethos away and you give that job to the models, what remains? And these are the harder questions, the ones that require a lot of discernment. And, and I think it's where people should spend a lot more time.

What's the UI on your Claude, your Claude bot?

The UI? I have a, I have a, I have an Autopilot UI that has like all my, all my tasks and like all my, my entire system. It's like just a little web UI, but it's almost like a Claude project. It's a— no, it's like a website, like a server that it runs. And then I talk to it on Telegram and I use Discord. For, uh, multi-channel and like sub-agent coordination and things like that.

So like if you're, if you're trying to set up a tennis match on Sunday at 3:00 PM, are you like, are you just, I would just send a voice note and then, and like, and like it would, uh, what if it needed to book the court and, um, it has, well, it actually has, uh, uh, something that we're shipping at Brex very soon.

Uh, it has access to a way of creating virtual cards on Brex through an MCP. So we go do that, go in the website, put a credit card on file, come back and say, hey, it's done.

So the agent would go do that. Is there like any way I'm spending 3 days assembling and answering questions about my personal taxes in 10 years? No way to do my tax return.

Absolutely no way.

Please tell me that's not going to—

Well, and it's going to happen on both sides, right? The IRS will also be doing your taxes for you. And then it's going to be— Tax preparer. Yes, exactly. So it's— I think that's probably true already to some degree. But, but, you know, I tell people everybody's going to be audited because the cost is going to be zero. Yeah, that would be an interesting future.

Like, this drives me insane. It's like all I do is a lot of hours just to gather up all these documents every year. It's unbelievable. And my, my computer could just be doing this all year. Unbelievable.

Yeah. Yeah. And, and the thing that's interesting is it's a fundamentally different problem. It's almost like there's the tax software and there's the bookkeeping job. And the bookkeeping job, part of it is interacting with the tax software, but another part is gathering out the documentation and going in and asking you about specific expenses and transactions and what they were and what they weren't. And yeah, the bookkeeping job is the really interesting part, which will become part of the software eventually.

So what, like, what, what, I mean, accountants become like lawyers. I mean, they're just handling the highest, most difficult questions.

I think so. I think so. And I think the, my, my, my thesis is like, for example, in accounting, I think you still have a bookkeeper because you need a CPA to file your taxes. That's like a legal requirement. Like if you're filing someone's taxes on their behalf, you have to be a CPA.. And the beautiful thing of the system and the way it works actually is a CPA has some degree of personal liability in case there's something wrong with your taxes. And that creates very, very good back pressure in the system where, well, because someone is personally liable, they better make sure that the data's correct. So they better make sure the tools they're using have a high enough bar for accuracy and control. So I think that's actually a really powerful thing where I, like, even if you had like Harvey and let's say the law, the law, you know, agentic AI, sorry, agentic law firms doing really great work, having a human in the end matters because, you know, we talk a lot about this idea, like what if Jim makes a very big mistake on hiring? Like, do you hold Jim accountable or do you hold an employee accountable? And actually you need the same oversight mechanisms that you have of employees. So Jim has a manager. There's someone on the recruiting team that's actually Jim's manager, so signs off on all the actions.

Has to have a hard talk with Jim.

Yeah, exactly, exactly. And give feedback, right? And give access to tools and set a budget and have ways of giving feedback and make sure the feedback's incorporated and fire Jim and change Jim in case things go wrong. So I mean, firing is more figure of speech, Ultimately, the idea, I think this idea of oversight over agents and how do you get humans comfortable with the fact that agents are doing things on their behalf, I think is a very fascinating problem.

Have you had an AI agent pay for something on your behalf?

Yeah.

Like through Brex or through the service?

Through Brex. Well, Open Claw through a Brex-initiated card, a Brex-sweated card.

Okay. Like you're doing this all the time or this is—

No, so I mean, I don't do it like I probably do like once a week, I would say, even though I make an online purchase probably like every 2 or 3 days. So I would say it's like a third, but I'm also trying to deliberately push the boundaries. So a fun one was we had a couple of friends that were going to the movies and then they sent us, hey, here are our seats. And then I literally just forwarded the message to my Open Claw and I said, buy the tickets. Nearby. And then it went online, went to the Cinemark website, found the movie room, found the time, found where they were seated and said, well, these two seats are taken. You should take the ones next. And then I was like, sure. And then it went online, put the card details, made the payment, and it was done. And then I got an email confirmation and I didn't touch anything.

Do you have a name for your open-class system?

I do. It's a funny— I call it Lemon Pie, which is a weird thing because when I was 12, my first AI agents, I call lemon pies my favorite dessert. Uh, and every sort of a bot that I do in my life is called Lemon Pie. So I was like, okay, this thing has to be called Lemon Pie. Uh, but, uh, anyway, it was a, it was a sort of funny historical accident.

Okay. Yeah. Okay. Okay. Well, I want that system. I, I look forward to you making all these things available in price for us to use. Yes.

I hate open source and in price.

I hate doing my expenses so much. Um, I've always hated that. It's pretty funny cuz I worked at the New York Times and Bloomberg for a long, very many years and their systems are so archaic. And when I did this startup and, and started encountering things like TaxAct, I was like, oh my God, thank, thank the heavens that I no longer have to toil at all these things that destroy my soul.

No one likes doing an expense report.

No, no. And, and I'm just, yeah. And I'm just, it's like self-defeating.

Exactly.

Cause I, I just don't do it then. Um, uh, well, thank you so much for coming in.

Thanks for, thanks for having me.

All the support. And, um, when we finish this, If you have a few minutes, I wanna show you a cool video.

Yeah, for sure.

That you guys are in. Uh, let's do it. All right. Uh, Pedro, thank you so much for coming and, um, we will see you guys again soon.

Yeah. Thanks for having me. Really appreciate it.

Thank you. Core Memory Podcast is hosted by me, Ashlee Vance, and/or Kylee Robison, or both of us together. It is produced by me and David Mickelson. Our theme song is by James Mercer and John Sortland, and the show is edited always by the John Sortland. Thank you so much to Brex in Ewan Ventures for all your support. And thank you most of all to everybody for listening or watching. We love you. Please leave us a like, a review, a subscribe, all those tremendous things. Thank you. And we'll see you again.