Final Advice for Future Founders

You wake up Whatever problem you have in your life, why can't you solve it AI? And just like start there. I think the CEO needs to be the chief AI officer. L it's not an engineering team thing. It's not like a product team thing. It's like You have to understand the bounds of a technology better than anyone. I think a good proxy for how to spend your time is what are things that only you can do in the models cannot do? You have to sort of refound The very concept of what the company self identity is Welcome back to another episode of The Light Cone. Today, we're joined by Pedro Francesi, co founder and CEO of BreEX. Pedro started BEX in the YC Winter seventeen batch and built it into one of the most important Fintech companies of the last decade. He's here today because BEX has gone deeper on AI than almost any enterprise company we know. And Pedro's own AI setup is so compelling that when he came to YC for lunch, it sent our entire team down a rabbit hole of building on their own Pedro, welcome to the Like Con Thanks for having me. Excited to be here. Thanks for changing our lives. Yeah. Oh God. that lunch I'm like, I think the model company should be sponsoring me for the token consumption increase I gener we supposedly generateated on that lunch. That was the precursor of G Brain, I guess. I was still working on GSack. I was still a twenty thirteen web two point zero engineer who time traveled instantly to the AI tools of january twenty twenty six you know, probably half a million lines of rails code in I' create a G stack because of that to like help me make a software factory. Yeah. And then after I met you, I realized everything is about freeing the claw. I kn you that. Yeah. And then and give it tokens. Yeah. Well, no, I mean, let it rip. The craziest thing was realizing like what I had gotten wrong that I think actually most people in software are still getting it wrong is you they've been treating the LLM like this very precious thing that's very expensive. And so as a result, you have to literally put the agent inside a Foxcon factory. And it's like like can you imagine? like, I mean, that's what the half a million lines of Rails code was for me. It's like, no, no, I need to control what the LLMC iss because it's about really, really I only want the context like from here and let me write like all the if statements to make sure like you know like a Foxcon engineer, you're waking up at six AM and you know if you don't, you're going to get electroros shhocked. I mean, like it's like this terrible thing that you do to agents. Yeah. And they want to be like at the EsLent Institute, and that's what openenclaw is Exactly Exact And it's funny because I feel like every single good AI product you've used is An Asent loop with tools. That's it Like there's no you try to sort of over engineer the harness and then do certain things, but at the end of the day, it's skills tos and Like there's not really much else. Maybe we start earlier because one of the things we love to kind of, you know, get down as a part of lore is like, how did you So AI pilled and like all the way to the edge. Well, I'll tell you my encounter with LMs, which was so I remember in the pandemic There was someone someone gave me an API access to GPC three And and I was playing with it and I was like Okay, this this is really cool. This is there's something here that could be could be special. But it was the kind of thing that was like, yeah, it feels like a research project, the kind of thing that Google used to release and you like you play with it for ten minutes and you stop. Chatu Pi came out and I think everybody was sort of interested in it. whereere I think it got interesting was when you started to see Reasoning models and of course tools, but but I think everything else was sort of a blip until December U and the way I describe it to to my team is like, You know, electricity was invented in December U And I think electricity was opus four point five. and sure opus models and you know, open eye models got better and better since then, but To me, that was the tip of the spear where you could say, yes, like coding harnesses actually work. And you know, Cloud code existed for probably a year before, but it wasn't that that valuable yet. And I remember, you know, during the holiday break, I was playing with it and it was pretty shocking, probably similar reaction that everybody here had. And I think the question becomes, you know, if you sort of if you think about you know,'re you're sort of standing, you know, looking at two hundred years of history And then you imagine you are, we're not May you're sort of five or six months after electricity was invented And most see' still playing of candles and you know, questioning, you know, what can you do with candles and fire and you know, it needs light Yeah exactly bright these lanterns and what can you do with it? And and you know, the steam engine is like, I don't know, maybe like twenty years away still. but you know, electricity already exists. That to me was the sort of the fundamental a light behind it. and I would say I think since then, open cllaw was kind of a interesting sort of next step, which is I think when we realizeed that you know, the reality is good AI products are Aigantic looubes with tools Uh and we started doing this in our own producta Racks, but but then on the personal side, I started spending a lot of time understanding, okay, what is at the frontier? of using openCaw and I think the insight was just yeah, like markdowns can take you really far. just like configuring and automating a lot of the things in your life. It's kind of funny. I rememember I had this this experience of like buying a movie ticket entirely in open clloth, using like a Bx card that was provisioned through an API And and then I showed it to my team and they were like Oh, but like you can go online and like book it in ten seconds and I'm like that's not the point. You'reissing you're completely missing the point. But anyway, and then I went obviously very deep into his rabbit hole, and I started spending a lot of time thinking how to change the fabric of the company and the way we build the products. And Tell us about your personal open do journey. Be I before you came for lunch, I had it like inst stall, but I was like way too scared to do anything with it were al. Yeah. Don't get me wrong. like we deal with financial services data. We spend a lot of time figuring out how to be mindful of security and protection, but and yet U, I think people are a little bit more risk averse than the technology probably requires them to be given where the technology is. And when we started using open Claw personally, I started doing it a lot of my own personal setup. Basically what I did the view one was I' want to give it read access to everything and just create like Of tokens to my email to Lack to everything to just literally not write. And I was kind of shocked how far it got me. Um, and then the next question that we we spent time on Backs was, okay, like How do we actually get it to right into our systems And everybody in Kooger and security team was, well, we cannot do that for all the reasons that we know. And then basically where I spent, I don't know, probably four weeks of my time was Okay, let's solve the hardest problem, which is security And we ended up realizing that the only way to actually do something about it was to something in the netwk layer. Uh and if you treat the agent like you know, the agent has Ps own wills, desires and you know, they go to the Asalent Institute for agents. And you know, they had a Fx Con factory Inent Fox Cons factactory, they will try to do things that are network boundary that that could not be the right ones Um, and we dec hadided to actually just focus on that So a lot of folks were, you know, and we saw Nvidia and others Nemo cllaw that's buildings like open sheelll forks that have controls over, you know, what tools the model calls. And the reality is Yeah, you can do all of that, but you can also just make an HTP request wrong. So we focus on that layer and then we build this thing called Crabtrap, which we open sourced probably about two months ago which is actually the way we use to secure agents at BX in production And that the basal prmise is You analyze your HTV proxy The entire network boundary of an agent And the idea is when a request goes through, that becomes auditible And you basically can use an anotherer agent to analyze the traffic. and create a policy to let traffic go through or not. And surprisingly or unsurprisingly, because these models are trained on you know, hundreds of billions of web documents. HTP traffic is actually I would say probably the way the models reason more so than anything else because they just literally learned on the web So the ability of the model to watch like thousand requests and make sense of what's happening. was way higher than we anticipated So so we actually build that, put that in production of Rex and afterfter you record a traffic of an agent operating for a day, you can build a pretty good policy uh that, you know, sets things that should be automatically approved. and for things that the agent isn't really sure. you can just use an LLM as a judge and the LM determines is this request something that should be approved or not based on the policy for what that agent should be doing. So for example, we have like a recruiting agent of Bxs called Jim. U We have a policy for gym and you know, all the traffic goes through that same policy and ninety eight percent requests go through automatically, two percent using LM So we sort of got that problem solved to a degree that ever got comfortable experimenting much more aggressively and sort of freeing the clot on the enterprise, which is really hard inside capapital One. So I wouldd say if we found a way to experiment with these things and granted, we don't do The most aggressive things with this stuff, yeah, we don't use it on like You know, customer data to the degree that we want one day to do and there's boundaries to how we do it. I don't see any reason why a YC company shouldn't be the bleeding edge of the stuff. Yeah, I mean, I think your intuition around like the proxy at the network level ended up being quite prescient. Like I think a lot of the stuff that I'm seeing kind of around the open cl ecosystem at the moment, at least Or just agent ecosystem is essentially doing that. Like we're seeing that like with credentials, credential broering, like aggent Vault is doing a all lot of that. I think you mentioned the first version of Crap Trap included, like credentials vote. Why did you decide not to include that? I think it was just let's just do one thing really well. And and, you know, at the end of the day I think there's going to be a lot of solutions that do that. You could do credential brokering and other tools already But the elements as the judge was for us a determining capability to say Do you trress us in Brush or not? But then our screwity team at Brs very rigorous and very good at what they do For a long time well, you know, not really, to getting them to it, yes, we actually believe this is enough was a big unlock for us. And look, I always say this like we're not in the business of building HT proxies We are in the business of bleeding into the bleeding edge of what we can do vI. And to get to the bleing edge required us to build this proxy. That's why we did it Hopefully someone's going to build a YC company,opefully we're going to build a better version, and we're just gonna to go use it But at the end of the day, That's the journey that took us to just sort of being at a bleeding edge in that way. And how much was you like sort of pushing this forward and they how much resistance did you get internally? and just how did you I mean AI peed? I think there was a lot of excitement about it, but the way I describe A adoption is how most companies is I think there's like Sort of Three tiers There's tier number one, which is your token maxers, like your engineers that are pushing a bunch of code and typically living inside coating harnesses And and those are sort of, well, now and we know who those are Then you have the the sort of average engineer that is building a few things, but you know, not Nasarto can master to the same degree and probably I't know tenth of the proivity. And then you have like the entire rest of the company And the entire rest of the company typically is interacting with AI in what I call like Google seearch modeway, which is a chatbot with a few MCPs U or a G Suite equivalent that yeah, you have a few tools from Google, but at the end of the day, it's really just like a search. And I think that our thesis was If you think about the value that AI creates for like a token maxer, for example. A lot of the value comes from the harness. And the thesis was how to actually build in equivalent harness for a our teams that are non technical. And our whole sort of thinking behind it was like, that's a lot of what openen cllog ced, which is ability that you can self would strap a lot of the capabilities of the agent by the way you edit your skills and markdowns and sort of set up the environment around the agent And how far can we get this ability for the agent to self boost travel capability without anyone actually going in coding it by hand So the analogy we use internally for I would say the sort of the company wide adoption of AI is We don't believe in the yes, like give people a few MCPs and let them go becausecause I think what people really want is, in my opinion is really a way of saying Okay, this is actually a virtual employee almost that has, you know, this on sllack, has an email. I can actually invite it to a meeting, can join a meeting, take notes And you're trying to replicate that as much as possible So how do you build infrastructure to support that kind of use case? And I think the harnesses will look a little different probably more like openen Claw than a Coe model Jared and I just did this this week for the first time where we installed AquaVoice. and then you open telegram with the claw. or actually we have it in Slack now And then basically it was like Me and Jared and like three engineers and are someone from the events team and we're trying to put together how do we put together sixty dinners with twenty people each of attendees from startu school ice with twenty one partners and visiting partners at YC. Sounds like a great problem. And then we just basically started talking about it, and then I picked that up and then I pressed enter. and then, you know, our cllaw just started doing it. None of us opened Clawud code. Like it just sort of built a bunch of markdown, it did the analysis. and Yeahah, people forget that Cloud code is magic is just literally a harness around the same models you can use in an API, right? So so I think that's the unlock of. And by way, there's a few things that Cloud Cod is doing that I think are really cool. Oh, they're amazing. And yes,s just it's just a harness. Clad can use Cloud code. Exly Codees, right. It really prefers to use Codex these days. Exactly really everything actually. I don't know why Exactly.. But ACP helps and not so. Yeah, ACP is good Why do you think the adoption of Token maxine hasn't really taken off? I thing that we found a very curious workking with a lot of startups early on is A lot of founders are very shy about burning tokens. I think you really get to experience this when you really go all the way. Gary mentioned this point, which is tokens are expensive And I think there are You know, I'm in a fortunate position to be able to spend on toens. but I would say I keep trying to picture myself. Imagine if I was like fourteen or twelve when I started coding for real. and I had the technology we have now, I would be token maxing in the cheapest way possible. And there are people doing that. you know, you look at the Chinese models, for example, like they're pretty decent. There's a huge hobbyist community where they you know build a gaming rig, but then they try to build like local LM. Yeah. And then is like totally reasonable way to do it. one hundred percent hundredcent. I have a friend that has the exact same setup. He has his like little Jpu farm in his house And and first time in there I was like Wow, heating's on here. It's really's hot in here. And he's like, no, it's my GPUs. And it wass like, great, like you know, power efficiency all the way through. It's funny because at Brax and we should talk about managing token costs and spend management for tokens, which is a topic we're spending much of cycles on now I think that the cost part is one But but even even if you take the cost spot aside You know, the first symptom is a lot more people should be complaining about the Max plan limits And, you know, you see what's the percentage of Twitter that probably complains about it? like point one percent. So so I I think more probably st early. to me, there's this like The AI pill test in my opinion is whatever problem shows up in your life Do you default to AI first or not? It's like, of course, mechanically, you can do it But there's a point that it becomes like second nature and then your whole like brain gets rewired and you cannot think in a different way And there's the whole topic about AI dependency, human machine interaction. Yeah there's all these things that we can we can talk about and put in the corner. It still sort of surprises me how many people you go talk to about a problem and I'm like It's so cheap to intimately understand the bounds of this problem now. like why haven't you done that yet and come in with like a much more digested view on the problem? And I think the second thing is like, think I think if you have the luxury of building a company now, The fabric of the company from day one can be built in such a different way that I think If I were to startort a company today, I would say Okay, the premise is, why can't he be just me Like and then you start from there and your token consumption is probably going to be a lot higher than if you said, well, I'm going to have like three people or five people or seven people And I think the fundamental constraint isn't as much, in my opinion Uh like oh, like AI as a cost savings so I'm going to be more efficient. I think the unlock is like the fabric of the company just looks very different. when the boundaries become tyype systemstems interfaces agents talking to each other versus Uh, and and I think people are still D didnn't fully grasp by, okay, what does it mean to build code with new agents, like and the new technologies we have, I think that's like well understood How to live in a world where Intelligence is on a tap And your default answer is Let me actually solve this problem with AI first. E even if you feel subopptimal And then from there saying, okay How do I actually make it optimal? Because I think for the majority of problems, there is a way to solve it AI that is probably better. And your job is to figure that out. evenven if it's going to take you more time because that will come down YC Startup School is back. We're hand selecting the most promising builders in the world and flying them out to San Francisco for july twenty fifth and twenty sixth to discuss the cutting edge of tech and startups. Apply now for your spot When you start Brecks, I mean, like it's well known like you're like MVP like had no Web UIRI which is like all terminal, like super scraft. Tod they would have, because Yeah, that's one' use HTMLs to test anymore. Like Was it actually still the right approach to justust have a really simple MVP and test that any more work wouldould you have like a way more fully featured? So so I have this controversial view, which maybe you all will disagree, which is like I actually think if I look into a pattern of companies that succeed I think there's a really interesting pattern, which is minimal surface area. And the problem is with AI, I think you see like look at Stripe, for example. Stripe earli days was like literally an API. Brexiter earays, no UI, just like literally a terminal Um, you look at Airbnb iss like the we the website was a form And the form was just like literally where you can put it what you needed. And then someone somehow went there and figured out how to actually make the booking happen. Like doorash in the early day is similar, right? Like it was just likeally so the surface area was so small with the customer And so much of the the band the sort of the intelligence and the bandwidth of the founders We're spent nailing this one single interaction pattern. And I think the risk with AI is that the agency behind choice goes away So so you have this, you have this this this lack of discipline on what matters to solve. And I think people tend to believe that I can just experiment a lot of things and that's absolutely true But that doesn't preclude you from actually choosing what matters I always tell people like I think if you don't, if you can't minimize your surface area and solve the problem with a very clear set of boundaries You haven't found a problem to solve And I think that's and you can of course findind how to compress the problem into a smaller surface area using AI, and that's really valuable But I don't think yourreoes as an excuse to not do that, which I think is why. I can just build so many other things But you know, I always tell this to people like intelligence is compression. So when someone comes to pitch me an idea in the company, I'm like, It has to fit in a napkin. Like where do just fit in a napk? What's a nkin? And then someone comes with this and I'm like, I don't know where you bu nins but the ones in my house are not this sizeed. How about the step before it, then A actually a lot of the pivot advice I give founders during the batch comes from you talking about how you found the brerex idea. And if like the approximate view I remember is like you thought about it as like two week cycles and like you're either in like exploration or exploitation mode and you're like trying a bunch of things, but then you want to like hone down. like would you still use that How would hundred percent? I think one of the most one of the hardest things of building a company is talking to customers and not not just having the conversation, but how to extract the sort of unspoken signal from these conversations. And I think to me the Can A I solve this lens? likeike whatever problem shows up in your life, can A I go solve that Any think about like building a successful company like, why can't you prompt your way into that And the reasons were simple is because there's signal that the models were in trin off. And the signal is when you go talk to person and they tell you about a problem they have They're not going to tell they're not going to give you the answer shaped into a prompt that it can put into an LM and that LM is going to go and output the product that's going to win and be a billion dollar company They're going to tell you a very sort of local optimum answer based on their worldviews and their constraints and the way they see if things And And I think a lot of the job is the job now is to have the wisdom to choose what you want. And because before the wisdom was not just to choose, was to choose and know how to execute it. The execution is out, right the execution is gone and the model is going to do that better The wisdom to choose is still, I think the the missing bottleneck. And to me that all comes from which signals are not in the models. So say like pre AI, you had personal bandwidth to explore like three ideas in parallel. You're saying like now in AI world, you'd still do three In parallel, would you like thirty and let the models try and The way I would probably approach it is But like let's let's biger broader universe of things to do sort of an early initial exploration. But to me, the lens is, okay, why can't I solve it? And which signal is not in the model And I think the signal is typically the customer. And then and then when you go talk to the customer, I think I wouldt paralyze that probably I would be, okay, let's try to get into headspace. of this person And and I think there's like it's so easy. and we did a lot of exploration with like synthetic customers and building customer role models and things like that. And and those are really valuable Once you know a lot about the customer But when you' don done though enough yets I think there's this like, very basic thing, which is even at Backs, for example, one of the hardest things for us as a company was We initially sold to founders, we were founders, we knew about ourselves, we knew about our problems. And then as the company got bigger, we were selling to finance teams. And finance teams were different So so building that mental model of like what's the value system? Like off course you can eventually make the model represent that and have that worldview. But but there's there's an intangible I think is is where a lot of the alpha still comes from. And I think to me is like the the I think a good proxy for how to spend your time is What are things that only you can do And even in the company of one, what are things that only you can do the models cannot do? And that to me is like one of them I think that's so on point. I think A lot of founders like you that successfully navigateated pivot this loop. Basically there's this book I authered in Mine from Psychology that has to do with people that have very good emotional connection with people are able to simulate what the other person is thinking and what they and others Theory mind. Exactly. And I think the founders I get that and have the empathy to figure out what the customer is not verbalizing is what is make the I think Gary saysays it explicit percent of what are all those desires? one hundred percent. And they're very subtle signs a lot of time because they're murmurs. as founders go through them and figure out the insights, o, is this really a thing? But how do you know went to poke for it Exactly. And the problem with relying on models and right now, which is I'm still very optimistic that there's still a lot of draftwered founders. definitely. is you don't even know what the right incantation or set of prompts to ask the model because you don't even know what to ask. Exactly. There's like another meta layer. Yes, it's the whole like elon thing of like, you know whichich question is the university answer for kind of? And of course, these these are generalities, right? But but I think what I've seen is you have to remember that LMs are not magic. L LMs are trained on a very specific coursese of information, optimizing for a very specific set of benchmarks and outcomes. And I think the biggest pitfall of LMs is You have no sense O how much training data The model has seen for the exact thing that you're asking it So imagine if like every time you ask an alama question. It gave you like, yeah, like I the sampling frequency of this in my Data set, was X and on this otherutter answer was zo point zero zero zero zero one x out of. You would trust is very different. right? The distribution is so different. Oh, I would pay for that. That's a great startup idea. Exactly. do that. We need to do a model that does that. Yeah. five percent. I would pay for it. Yeah Well, because it's fascinating because then like anything that's out of distribution, you just go and like fill that in. I mean actually as an applications engineer on top of the LMs, that's actually a huge blind spot. And that's what MerCore and a lot of the outut of data companies are doing. like a lot of the jobs for them is to say, well, what are the blind spots for LLMs? And And it's funny, like I think a lot of the data labeling companies right now trying to understand the bitfalls in the models But the problem is in order to do that, you have to be an expert to know what the gaps are in the answers. But the problem as a founder when you're looking for an idea is you know nothing about it So so the so there there is a there is a curse of knowledge and a curse of not even knowing what the bounds of knowledge is which I think can can can make you believe that you understand something that you actually You or a model, actually understand. Can I confess something weird about like after creating Gbrain? Now, I do use AI in a different way where Now that I have a retrieval system that is actually usable, if I have a problem or question about anything, h like for instance, I was trying to work on a really, really like The last humanized prompt. And, you know, a lot of that stuff probably isn't in distribution yet. There's a whole Wikip a Wikipedia article about like, you know, characteristics of u AI writing Now I can just go tell it. like go spend a day, like deep research literally every single paper, article, like read everything put it into my Git repo and then I'll be able to retrieve it and summarize it into something that actually is usable. And so it's sort of like filling in one hundred percent the things that are out of distribution.ike I can sort of like pack it with whatever context. And it's like you can do that with anything. It's like if you're interested in, you running a restaurant, literally you could have You could go and read like five hundred books about like every the top five hundred books about what it's like to run a restaurant. And you would have like the compendium of all information about it. Yeah. And I think a lot of what like for example, like one of the things that we do at Braacks now is building this customer world model is sim idea where We're trying to get every single touch point that the customer has of us, like literally like what How many times they click a button to the dashboard all the way to what they tell someone on an email or would they say on the phone or to send a call and ingest that one's holidata Okay, what should this customer need next from us? What should this customer be thinking about? Like what are the issues that they will face but haven't faced? And again, it's just a distribution problem. This is actually an answer to one of the questions, which is like, will there be jobs or whatever? It's like as long as there are limits on Uh RAM Actually like there will be. So I don't know. I mean, ' kind of an interesting one, right? I think so. Literally, you can't have a model that has enough parameters that could have everything that you could possibly need in distribution. Like there aren't enough atoms in the universe, right? It's like a modeling problem I think we forget that the world models in which the models are trained. there is something that the designers of the models influence the way the model actually behaves in the end. So so you know, one of the things that we spend a lot of time thinking is like, how to make LLMs work point look very different from us. How do we make ellens work for like average finance person in the U.S. that for're talking about an answer and you know, the model defaults to like AI CapEx as a fin as a default category for like Like for example, that's a really funny example. I was playing with AI for Cy categorization And like the first example of like an example of an is just like writing pros. and an example is like, AI CapEx. And I'm like, oh, why is it AI CapEx? the first example it comes up with? Because The people that are building the models fucking only think about it ACapx. Right? So there are things like that that I think is like kind of interesting to think about that the mental models of the model, I think are Out of the box are more biased than we may give them credit for. I mean, spepeaking of AI CapEx, like earlier you're saying, you know, we're so early still. I don't know The funniest thing about AI to me is how often I find myself thinking crypto maxims. Yes. This is the worst The models will ever be. Yes. My favorite now is telling people who hate AI coding, like have fun coating at one X speed. Exactly Exactly. Eactly. I was telling a friend about, you know how how how to be, you know, long inference that basically the thesis is that there's going to be a lot more inferenceces that people think. And people are expecting a lot of inference if you just look at public markets and you, semi supply chain, all that. People are saying like ten thousand X But but the underwriting, which is kind of funny is like I think there's one image. So thousand five hundred dots, each dot is a three point two million people on the planet And basically, you know, eighty four percent of the world never used AI. sixixteen percent have used at least once a free chat bots. And then zero point three percent, which is, I guess six or seven squares pay twenty bucks a month for Ii And one box out of the twenty five hundred actually use agents in whatever capacity So And that's the argument to be long inference and I think I think it's just is just starting out. And I think a funny thing on this is I think the It will be the biggest expense in a company like easily, right? And And yes, there's a lot of margin in tokens right now, but People always want to be at the bleeding edge, but even token costs decrease by ten X. They're going to have toX more usage, so it would be still a large cost Um, and we're spending a lot of time thinking how to help companies actually manage Token spend on Bs. We We ended up building our internal version of this. We call it Mag buy where the idea is it can effectively you know, every dollar of token spent in the company you can attribute to a product we have to customers, an internal tool that we use to serve or an internal employee uh, and understand model, usage, et cetera. And we're now figuring out how to do the analytics on What are we trying to do with the tokens Um to start to get a sense of ROI U But anyway, it's a fascinating topic that I think has a lot of really Ely early work compared to what it will be one day. Can you share any of the data that you've gotten from Brexit about just like what token spend is like in the economy It's increasing. No, look, I think two things are surprising. One is I think to your point earlier on how do we look at token maxing? I do think there' such a thing as how much cost boundaries you create internally dictate token consumption, obviously. But to me, I think what's the most fascinating is When you look into the sort of time I rA is written now Uh and may of you include New York. U tons of token consumption and you could probably argue, and we've se in the data also faster revenue growth I think what's really interesting is the gap between Anyone in this two H mile radius and everythingthing else. And this is like not small companies. L you look into like very large companies with very large budgets And that could be token maxing And the economic thing for them to do would be token Max and they spend like I't ten thousand month. And you're like, you should probably be spending ten times more or twenty times more or a hundred times more. That's still surprising And and I think the reason is again, sort of similar to the point in the beginning. like we did this exercise two and a half years ago where I sat down with you know a lot of the engineering product leaders in the company and we had this question, which was if we started Brex again in twenty twenty four The answer would be even more different now. What would we do differently And turns out like everything. And they start going down this route and it's like it's kind of maddening because they're like, okay, we have this like completely old way of like even thinking about the fabric of the company and and the way we build the product and the way we build our processes internally, the The first best answer is yes, we wish we had started now. secondecond best answer is like let's go do something about it and change the way we do things, right? And I think a lot of our approach in terms of adopting I has also been You know, how do you How do you pause and say Okay, like there is a discontinuity in the not just in how we solve the problem, but on what the definition of the problem actually even is and sort of take a step back and rethink it Um, and, you know, like there's there's like millions of examples of that, but you know, one example, which is kind of funny is You know, we're redesigning KYC process. Like whenever we on board a customer we have to do all these checks to KYC to the customer And KYC historically is something that you can automate like eighty percent of it twentyty percent is manual And of course, the original impetus for anyone is this' with a nation that does it Yes, we can go do that. But what we decided to do is actually say, let's redesign the entire process into it And then what we re desesign is the entire onboarding process And when you redesign the entirebarding process What you realize is there's a very important thing that happens in the beginning of the funnel, which is deal qualification. Like is this customer even remotely qualified to be a Bx customer? But when you have KYC for free. You can you can KYC a lead versus a customer So you start to have risk orientation up in your funnel and that changes who you even target because you know who's going to qualify and the same thing is true for credit to some degree. So now the bounds of the problem have changed And And you can go in and say and I think a lot of including a lot of our competitors had this approach of saying Oh, I have this entire old process. Let me go and like latch on AI on top of it or lastash one A eye on top of our product. and I think the biggest iscuities in a positive way that we've had where when we said, hey Let's keep this old way here put it in a corner and like how do we design it if for start of the company today from scratch? and then just doing that. It takes a little bit of founder energy to do that, but I think' the It's the only thing we've seen working to really sort of inflect. I think that reminds me a lot about this is sort of a way back, I don't know, if you ever try to compile arc distributions of Linux Nure within power uses of Arc Linux versus Ubuntu is very different. Very different. I think the Ubuntu people kind of feel more like people that try Chati Piti. Stuff kind of just works out of the box or some stuff that you can get up and running. There's still not a lot of people that use Linux by the way, which I think it feels where AI is, but with Arc, you're like super hardcore And I think that's what open cllaw and herermes feel like because you have to really customize it to your own use case Maintain your skills. You have all the markdowns. and if you get it working, you can build something awesome. One of the most impressive things I've seen people build with Arc is actually, I don't know if you know Valve, the steam engine The operating system. that runs that makes it feel like a Nintendo switch is actually built on top of arrc interesting. They customize all the drivers over the air updates. It works with all consoles, it works with all sorts of hardware out of box but they super duper customizeed it. And I think this is kind of what's happening If you get your open clw will work really well for you You can kind of build your own custom Nintendo swwitch for whatever you need to do. Yeah. I always have this thing that I tell people, is which is funny, which is Think about your times to years ago I feel like you're working a lot more now than two years ago, right and pricing for everybody here. So then the argument is, well what about the productivity? Where's the productivity? Right? And And I was talking to Sle a very large public company this week And she was telling me that we see all the so and consumption and And you know, we're trying to measure like product velocity. And we're seeing like more lines of code pushed So So yes, maybe that's the way to measure the ROI, but Is it really there because people are spending so much on tokens And and I think the I think I think this analysis, like, yes, of course I think having sense on RI on tokens is important But I think it misses the point You're standing in the timeline of history And it's six months after electricity was invented thinking about like imagine someone saying in like I don't know, eighteen the eighteen hundreds like, oh my electricity bill is so high now Like gosh, let's use a little less. Let's keep let's push this steam engine to come like maybe twenty years later because the cost savings Yes, of course, don't bankrupt your company and tokens. It's actually a perfect analysy because I don't know if you know this, but when electricity was first invented, it didn't work very well and the ROI was actually bad. And so if shortly after the invention of electricity, some of accountants had done this analysis, they would have been like, o, this electricity thing is like isn't never going to be a thing. The ROI sucks Why do people stick to it And and It wasn't it wasn't cost savings. It was just because people were curious about it. And I think I think the point of like Why, you know, like I was yesterday until two AM playing with slash workflows and opus four point eight and all that is because I think I'd do the exact same thing if I wasn't making any money because you just see the possibilities and you see what it can do to technology. That just drives people to behave differently. and I think That to me is the ultimate lit mest and And it's a good separator. and sure if tokens are so expensive, they're going to be I think Over to Fs of time probably free Uh if a project is, I don't know, a hundred years online. almost compared to what electricity Now we don't think about electricity costs in our dataays, but unless we're in a dataenter But but I think there's something there's something similar for sure. We talk to a lot of founders of later stage companies H wish that their companies could be like as AI pilled as possible. And you run this like big company now with all of these employees. and that's only the Bck side. There's also the Like How to said, I'm curious, what you've done to like bring the rest of the company along with you on this journey and if you have advice for other people other CEO. There's a lot to do. I think the CEO needs to be the Chief FI officer. L it's not a Engineering team thing. It's not like a product team thing. It's like you have to understand the bounds of a technology better than anyone. I would argue that uness you unless you really experience the limits of technology every day, I think it's really hard to even understand what it can possibly do. Oh, you know why? It's because nobody can say no to the CEO except the board and the board being in the weeds per se. That is one hundred percent true. When you go think about you know the whole the whole example of KYC that we were saying Like the KYC team would never think of using the KYC technology to score a lead The only people that can think about organization of the system itself is if you have the context of the whole. And and to me like the The single most important question that NCO needs to answer is forget about the competitive landscape. Imagine you could getet the state of the technology today and transports to the moment you started your company. The opportunity was still the same But just the possibilities of the way to build a company are totally different How would you do it And then diff this versus what you have And then first suffer in silence for a little bit because you will, I mean, I do every day But then the second thing is, okay, what do you do about it? And how would you do it if you were starting from scratch? You would be the one figuring out Okay, how do we design our on boarding process or how we design our growth engine and our customer acquisition and the way we talk to users and the way we synthesize the data. All of that be would be redesigned from scratch So I think it's like it's almost like you have to sort of refound The very concept of what the company's self identity is and the way the functions and people's sense of success get structured. AI is an umbrella that I think has like three things the way we talk about it internally. There's product the product we actuallyers shhip to customers There's operational AI, which is things that directly affect our ability to serve customers at scale like think of customer success, risk onbard operations, et cetera. And there's sc corpor pretty high, which is how people work internally. The three agendas matter. And they matter in different ways, depending on the timing of the company Um, and and I think people will sometimes sort of pion hold themselves in one of the three. But in reality, I think you have to take a step back and be like you know, the same same thing we're talking about earlier, like, why can't you solve everything of AI? at a limit, that's the question And then sort of start from there and start problemsl around that question. It's a turnaround, almost. I think you have to assume that if you're a big large company that's not II native, you're doing a turnaround Uh to some degree. I guess we've been making fun of Fox Con factories for some time. On the other hand, like if you look at them, they're like this paragon of like very extreme efficiency Yeah. but they also are designed to be that to like create one thing perfectly back to back to back. And so you have to build a factory like that. And most companies are designed that way.ight. I think like processes are designed not to change. Yeah. There is a certain amount of broken glass required. the question is ike I think it's ten X easier for the CO to break last than an executive. And ten' easier for an executive than an employee. So you know, a lot of times like someone comes to me and says, I'm trying to do this AI, but someone is saying no because we haven't tested this in this use case or in that thing. And I'm like Okay, what are you trying to do? like Do you understand the risks? Do you understand the guardrails? Yes, okay. It takes me literally ten seconds to solve that problem It would take someone ten hours to go in into the meetings and escalate and understand Okay, can webe say orr maybe never, never. And I think the conclusion is probably never because most people would say, You know what? I'm just gonna like build this product in the old way because like why wouldn't we it just works? We know it's that guy's gonna hate me and then I have to look that person in the lunchline every day And it's like I want people to be happy and like me. So I'm justm not gonna do that. And what I tell people is think I think the escalation paths need to be like desensitized in the system because the company builds antibodies against any sort of Disturbance to the social cohesion of the company typically gets like rejected by the antibodies. And I think making escalations faster and being like, hey, We're gonna to go try this sing. you know, I understand the risks let's take this risk because the biggest risk is not taking that. is just literally missing the opportunity to rethink problem from What would you do if you start of the compereany today? On the corporate AI sort of leg of that store specifically, like, you buy into sort like the Jack Dorsy view of every company is essentially trying to like build its own little company, AGI, I do, but maybe in a slightly different way. I do think theomain specificity matters So like I don't believe in the like, I'm going to have like a single company model that has like every piece of data like in a single like with no judgment or lens into anything. So and the way I think about it more is like is more as the sort of the the virtual employee analogy, so to speak, which is like, how do I build an agent or version employee that is exceptional at understanding everything that matters about this customer. That is a well defined problem with clear boundaries of like clear APIs of people who who depends on the data, who interacted the data that is self contained then there's a moder agent that can be Okay, given all the customers that we have and the problems we have How do I manage my product ro nowap That can be a separate agent, but that builds on top of this customer world like a virtual exc team, basically. Exactly. Functional and domain knowledge still matter, right? These things are not going to go away. and I think the The way knowledge is structured, I think is still true, right? It doesn't necessarily change that much. and you should separate the agent and the systems that are actually emitting code from the system that is talking to customers and the system that is reasoning about the conversations with customers and translating into product of that basically three separate things. We're kind of the Tesla for AI. We're like I don't believe in anything that doesn't have real usage. So it's like, yeah, I build this great model and I'm like, okay, how many people are using it? Is it actually displacing the need to hire a person inside a company? Is it actually displacing the need to you know, spend literally hours. like how many hours is this thing saving? And I think a lot of times people say, well, know it's it's cool model. and I'm that's not going to cut it, right? Once you have that orientation, I think, customer ro model, okay, like your for example, our client saalle seam now runs on our customer world model.. So I know it works. I'm actually having lunch of a customer tomorrow. And I don't know the state of that account as well as I probably should customer model answered a question for me. and I now have a report with including things that the team didn't know about that came through support tickets and you know, an executive that was traveling had an issue at an airport with their car toot all the information awareness. Total information awareness, right That is a well defined problem. that is working. I can trust this building block as part of my company model as a whole. and you can have evalves on it. like we know like you know, I think a very we've stuck about EVols, there's a bunch of learnings on this and how to build EVols into the fabric of the company. but But anyway, I think it's more of like you have to decompose a problem a little bit Yeah, my favorite thing about EVLs is just running cross modal EVLs against each other. So one of the things that We' doing that. it is related, but I think it's really fun which is How do you have every single human interaction in the company becoming an evolved. when you have any agents. So for example, we have you onboarding, on boarding agents doing something. And then you have a team that actually goes in and looks at KYC exceptions that the model can't figure out How to make that a breaking change? And okay, like this manual interaction will become an Ealques. You know, we have an expense agent in Braxs Whenever someone has a conversation with the agents that is that flags an issue or a bug or something that feels like the conversation didn't go out smoothly. That creates a bug That bug triggers a nation that's going to go and modify the code base and the prompts and everything to make that evol pass And if that doesn't break, then engineer is going to go in and figure out how to make that ass becausecause the goal at the end, I think, is to make the whole thing a self a self learning system, right? And I think the A lot of what I see with companies is they spend a lot of time getting in Asian working but never thinking how to make the agent improve every day And I think that's like always the biggest unlock. You need a dream cycle You need a dream cycle see everything every night. Exactly. And it's like, oh, what's going on there? I need to put this over here. Exactly. What actually happened? Is there a pattern? How do I repause this? So how to bake the dream cycle into the products and into the agents and into things to ship. My favorite thing right now is I'm building like three or four agents for my friends Oh interesting. And some of it is like this is a user research for me for GBrain because it's like I have one. it's working really well. I have three hundred fifty thousand down pages in there now. What a crazy like I thought it was this wild you know, pie in the sky thing and it's like it happen' going to happen in our lifetimes. You know, I remember when when Neural Link came out And I used to think about it. I was like, I don't get it Like I was like, yeah, of course I get it conceptually, but why is it a thing And then now you use I and you're like, Yeah, ye, makes sense. Makes sense. I'm the botack. Yeah. Typing is so slow. I don't know if you use a lot of adaptation. I use a lot. My most used developer deeveloper UI right now is like voice memos to open clock. I've se this before, like it was maybe accidental, but I actually just really love the fact that like telegram works so well with with all because it's forced me to just put more stuff like make the agent more intelligent so that you can do more stuff via voice memos because you have to sort of fight the natural instinct as like a traditional developer where you're like, oh, like I can't quite do this, so I doesn't do this so I need to go like build more client or more UI or like more functionality for it. Foxc. exxactly.. just let it do what it wants to do. give it some context and it'll just think about, you know, oh, like actually, what about this? I think a lot of the work to your point is the How do you organize a context for a model and you can use a model to help, but but that is that is that is the that is the bottleneck for. Once you have the context in there, it's actually you can do some pretty crazy stuff. like my favorite new feure I I saw yourin LSD. Yeah. Yeah. Lateral synapacttic drift. So you just bumped the temperature on the surarch. It's not just that. So you have the vectors, right? Yeah. And so you know, if you think about what conventional ideas are, like most people give you like, oh, well, this idea, with this idea and it's like kind of like in this cone LSD mode actually says you cannot combine concepts that are withines. They actually must be orthogonal or just like like feeling seemingly random And then it'll try like, you know randomly hundreds of these combinations And then it'll rank order them into the ones that are actually the most coherent. And then if you do like one hundred of them, actually like the top five tend to be banger tweets. You know what's crazy is I didn't tell Alfred Exressly to be dry. I went I actually had like chat GPT generate like the soul file. It was like based on everything you know about me, all the interactions here, like generate like a soul MD for my open c agent and it was so unerringly like accurate about like kind of what I would want from like an agent. I was like, oh damn, these models know a lot about us My open clock got really interesting once I I just ingested my sixty gig I mean, you have to write a bunch of Hiku code to like only get the emails that are actually real, but you know, there's like extract like four thousand emails out of m gigs that actually matter. But like those are like, oh, actually like a lot of your thinking and you know the consequential moments of your life. So Pedro, thank you so much for being with us. I mean, you're by far one of the most AI pilled, farthest out on the edge, but also to call CEOs who is, you know, playing with this stuff and actually building it yourself, what would you say to people watching who are founders who want to be founders? You know I think that you are sort of the model for the way people should start companies and run them with AI as your SLN bud. I really can't stop thinking about the electricity analogy, which is, you know, you're standing There's a two hundred year timeline of human history. There's a point in time where electricity was invented. Its sucked in the beginning of or six months after at that point What do you do differently? knowing everything that would be true about electricity knowing that data centers when they consume electricity and even AI, right Well, you do a lot of things differently, I think. So I think that thiss one of just just marveling at the possibility of the exact moment in time where now I think the second is like, have a post it on your computer which is You wake up Whatever problem you have in your life, why can't you solve it AI? And just like start there in eighty percent, yeah, you can use a chat bot But at twenty percent that you can't figure out why and go build something that makes you solve that problem Less so because of the because of the immediate usefulness that sovereignnty thing at scale will have becausecause it gives you a texture and a feel for the possibilities for technology. are really hard if you're not playing with it every day. And maybe the third thing is like I think it's like just measure your toten consumption and how much you're just pushing the limits of the company and starting with the premise of like, okay, why can't you just be one person Like why can't it just be me that builds the whole thing? And they're going to probably face a wall of the the elements of, you know, what models can and cannot do, but at a limit, I think the question is U you know, how do you spend your time on the things that only you can do as a founder And these things to me are Number one, which problems are worth solving And two and the sort of the choice thing we talked about And the second thing is Okay, given that given these choices, what are the limitations of an outline that they still cannot do and I have to go in and do those things myself. but almost u you know, to some degree, you're working for the LM to some point. And and if you're in a bigger company, you're in to turn aroundound pretty all I' as almost a founder a CEO and you're almost architecting the entire company around that idea. But I think early on uh, so much of it is, you know choosing what matters talking to customers, injecting the signal that the models don't have, and just, you know Rebuilding is the way you would do it in twenty twenty six with electricity being six months old Thanks, Pedro.'s awesome. Yeah. Thanks for having me. Appreciate it coming