1% better

Why Now Is The Golden Age of Building | Brian Armstrong, Coinbase

How does a founder stay motivated by imperfection rather than paralyzed by it, and what does it look like to run a company with a team of AI agents?

1h 0m

Summary published by , updated .

David Senra

Key Takeaway

Write down how you want to think and feel about something difficult, and repeat it regularly.

If you dread conflict, write "I love conflict" each morning for a few weeks. If you want to stay focused on a big goal, write it down daily. The idea is to keep it top of mind so you start noticing opportunities and shifting your response over time.

Brian Armstrong's explanation is straightforward: repetition updates the pattern your brain defaults to, the same way retraining any system changes its outputs. This is less mystical than it sounds and more mechanical than most people expect.

Episode Overview

How does a founder stay motivated by imperfection rather than paralyzed by it, and what does it look like to run a company with a team of AI agents?

Brian Armstrong, CEO of Coinbase, joined David Senra for a wide-ranging conversation that moves from personal habits to company architecture to the future of finance.

  • Self-programming and fulfillment. Armstrong describes writing down beliefs he wants to internalize, distinguishing between fleeting happiness and the deeper satisfaction of doing meaningful, difficult work.
  • Agentic work at Coinbase. Armstrong walks through his current setup: lead agents for each area of focus, a chief of staff agent overseeing them, and a prioritized inbox replacing walls of chat text.
  • The financial system opportunity. Despite years of fintech growth, traditional finance still holds roughly 96% of global financial services revenue. Armstrong sees AI agents as the catalyst that finally gives people a reason to switch.

Coinbase Brain refers to an internal knowledge system that distills lessons from meetings, code reviews, and documents to give agents useful context when working on company tasks.

Main Insights

1. Repetition can change how you respond to situations you currently avoid

Armstrong describes being conflict-averse early in his CEO tenure, which made hard conversations feel draining. His solution was deliberate and repetitive: he wrote "I love conflict" regularly until his default response began to shift.

The mechanism he describes is straightforward. Repeating something keeps it top of mind, which changes what you notice and how you interpret situations when they arise. This is not about pretending a feeling is true; it is about updating the pattern your brain reaches for first.

Useful questions to ask:

  • Is there a situation I consistently avoid or handle worse than I want to?
  • What would I need to believe about that situation to respond more effectively?

Armstrong applied the same approach to his morning routine, consciously choosing what he repeats to himself before the day starts.

2. Fulfillment and happiness are different targets, and they require different choices

Armstrong draws a clear distinction. Happiness, in his framing, is temporary and tied to how a moment feels. Fulfillment comes from doing meaningful work, building something over time, and contributing to something larger than yourself.

The implication is that optimizing for comfort will not produce fulfillment. Armstrong compares building a company to playing a video game on easy mode: if there is no challenge, there is no satisfaction in finishing.

He is careful to say this is his own orientation, not a universal claim. But the practical question it raises is worth sitting with: when you are deciding how to spend your time, are you choosing what will feel good now or what will feel meaningful over the long arc?

3. Being perpetually dissatisfied can be motivating rather than demoralizing, depending on how you direct it

Armstrong says he is constantly noticing what is broken in the Coinbase app and pinging 30 to 40 people a day about it. He calls this a form of competitive obsession, and he compares it to building in a strategy game: collect resources, build, expand, repeat.

The important distinction he makes is that the dissatisfaction does not turn into a negative internal monologue. He uses the frustration as a signal that something needs to move, not as evidence that he or the team has failed.

David Ogilvy's phrase "divinely discontent" comes up here as a description of people who produce exceptional work. Armstrong treats the restlessness as fuel rather than a problem to resolve.

4. AI agents work best when given an area of responsibility, not just a prompt

Armstrong describes two modes of using AI at Coinbase. The first is the familiar one: prompt, get a response, move on. The second is what his team has shifted toward: assign an agent an area of focus, let it investigate and work, and have it return with prioritized recommendations for human review.

In practice, Armstrong has seven or eight lead agents running in parallel, each with sub-agents handling specific tasks. A chief of staff agent oversees them. The human's job becomes reviewing a prioritized inbox of decisions, not managing walls of text.

The agents he describes operate at roughly a mid-senior employee level on well-defined tasks, though Armstrong notes they still miss the big picture at times and improve steadily as the underlying models improve.

5. Internal knowledge compounds when it is captured and made available to agents

One of the more concrete things Armstrong describes is Coinbase Brain: an internal system that ingests meeting transcripts, code reviews, incident reports, and policy documents, then distills them nightly into a knowledge base.

When an agent starts work in a given area, it draws on that brain to produce output that is already aligned with how the company operates. The practical effect is that agents need less back-and-forth correction because they have context a generic model would not.

Armstrong frames the main job of humans in this setup as curating and aligning what is in the brains, resolving contradictions, and making final decisions. The system formalizes institutional knowledge rather than letting it stay locked in individual people's heads.

Frameworks or Models

Tiered AI agent structure

Armstrong describes how he has organized AI agents at Coinbase into a hierarchy with distinct roles.

  1. Lead agents. One per area of focus, each given a goal or responsibility rather than a single prompt. They investigate, analyze, and return with prioritized recommendations.
  2. Sub-agents. Each lead can spin up three to five sub-agents to handle specific tasks such as writing code, testing it, or coordinating with team members.
  3. Chief of staff agent. Oversees the lead agents, helping track priorities across all active areas.
  4. Prioritized inbox. A human-facing queue that surfaces the decisions most likely to unblock downstream work, allowing quick yes or no responses via voice or keyboard.

The human's role in this structure is to align the agents, review recommendations, and make final decisions rather than to manage individual tasks.

Coinbase Brain knowledge system

A nightly knowledge distillation process that gives AI agents company-specific context.

  1. Ingestion. Meeting transcripts, code reviews, incident reports, A/B test results, and policy documents are fed into the system continuously.
  2. Nightly distillation. The system reviews new material and updates a structured knowledge base, one per employee, per team, and per code repository.
  3. Context injection. When an agent begins work in a given area, the relevant brain is loaded into its context window.
  4. Human curation. Employees review the brain for contradictions or outdated information and resolve them, keeping the knowledge base accurate.

The goal is to reduce the number of correction cycles needed before an agent produces output that matches company standards.

Notable Quotes

"I want to be like embarrassingly passionate about what I'm doing."

David Senra

"The point of the game of life is not happiness, and it's not comfort or anything like that. It's, it actually should be difficult. It's like a video game. If it was like easy, it'd be boring."

Brian Armstrong

Action Items

Suggested exercises inspired by the conversation.

  • 1
    Rewrite a belief you want to internalize

    Pick one situation you consistently handle worse than you would like, whether that is conflict, public criticism, or uncertainty.

    1. Write a short, present-tense statement describing how you want to respond to it.
    2. Write it down each morning for two to three weeks.
    3. After two weeks, notice whether your default reaction in that situation has shifted.

    Armstrong's point is that the value is not mystical; it is simply keeping the intention top of mind so you start noticing and acting on opportunities to practice it.

  • 2
    Distinguish your dissatisfaction from self-criticism

    The next time you notice something is not good enough in your work, pause before reacting.

    • Ask: is this frustration pointing at something specific that can improve?
    • Ask: am I criticizing the work, or am I criticizing myself or the team?

    Armstrong describes using the frustration as a directional signal, not a verdict. The goal is to convert the energy into a concrete next step rather than letting it become background noise or a negative internal loop.

  • 3
    Map your work onto a longer arc

    Armstrong uses strategy games as a mental model for how progress compounds over time. Try applying the same frame to a current project.

    1. Write down where you are now.
    2. Write down what the next two or three meaningful steps look like, not the entire path.
    3. Identify one dependency or bottleneck that, if resolved, would unblock the most downstream work.

    This is useful when day-to-day frustration makes it hard to see that incremental progress is still progress.

  • 4
    Assign your AI tool an area of responsibility instead of a single task

    If you currently use AI by prompting it for individual responses, try giving it an ongoing focus area.

    1. Define a goal or area you want it to track, such as improving a specific product metric or monitoring a category of feedback.
    2. Ask it to come back to you with prioritized recommendations rather than waiting for you to prompt it.
    3. Review its suggestions and approve or redirect, keeping the final decisions yours.

    Armstrong's experience is that this shift from reactive prompting to delegated responsibility changes the quality and usefulness of what you get back.

Full Transcript

Read or copy the complete episode transcript.

I wasn't expecting to start here, but we were just talking before we started recording, and we were talking about brainwashing ourselves into behavioral change, things that we want to change about ourselves. We were talking about when I had Toby Luke on the show, and he said he used to be terrified of public speaking. And he would write down, I think, every day for like 10 minutes that he loves public speaking, and now he loves public speaking. He's talking about you can literally change the way you think and feel. And you're like, yeah, I resonate with that. I think he's right. I mean, I've definitely experienced this in my own life where, you know, I'll write down something that's like a goal or something I'm trying to work on. And I think some people have this like kind of woo-woo like thing about it, they're like, oh, you're manifesting something in the universe. I think it's actually much simpler than that, which is just if you repeat something often enough, it's top of mind. And so during the day, you start to notice opportunities that might help you move a little closer to that or practice it. And it's just a way of keeping things top of mind. I guess another mental model of this would be like, you know, your brain is like a neural network. And every time you reinforce something, it just updates the weights in your model, kind of like an AI model. And so literally just repetition, I mean, people have known this for thousands of years, right? Like prayer and meditation. And repetition just strengthens the neural pathway. I don't know if that's like accurate from a neuroscience point of view, but it's sort of a metaphor of how I think about it. So some of the examples for me, I mean, in my 20s, I remember writing down like, I want to start a billion dollar tech company. And I didn't even have any business writing that down because I'd never started a million dollar tech company or a hundred dollar tech company. But I wrote that down a lot. And I was like, okay, I was starting to look for ideas and opportunities and try things. Some of them work, some of them didn't. And you know, then I moved to Silicon Valley when I was maybe 29. And within six or seven years, I was the founder of a billion dollar tech company, right? So that was interesting. And then early on as being a CEO, I think I was pretty conflict averse, right? And sometimes being a CEO, you have to have just hard conversations with people all the time, right? It's like, these two people aren't getting along. You got to fire this one. You've got to go make a hard call. And so who do I want to disappoint today? Because only the biggest problems come to you, hopefully the rest are getting delegated. And I started writing this down. At one point, I was like, I need to reprogram my brain to not think of this as super stressful, but think of it as it's me getting valuable things done to move forward towards the goal. And so I remember writing this down a lot. I was like, I love conflict. I love conflict. I love conflict, right? And it doesn't mean I go out and like actively seek that and just to piss people off or whatever. But if it does happen, I can get somewhat excited about it now. I'm like, okay, I'm about to do something difficult and meaningful and like, let's move it forward. Or if I'm in a meeting with the team and I'm like, the work is not good enough. There's something wrong here. Like I'm feeling myself getting angry. I'm like, how can I use this as an empowered anger to actually help create change that will create a better outcome? So anyway, those are some things I've written down and practiced. Let's say you're in a product review meeting and you think the work is not like high quality enough. You would previously, you would suppress that anger. Yeah. I think I was like afraid of big emotions and saying something I would regret. And so I was kind of like bottling it all up inside. People could kind of tell sometimes too. And I'd be like, okay, guys, I think that this could be a little better, you know? And I sort of became more comfortable with it over time. Right. And I was like, look, this isn't good enough. It's not meeting the bar. Maybe I'm going to change it by the time I meet with you tomorrow. Or maybe I'm going to put you, you're not the right person to work at this company or whatever it is. Right. And people sometimes afterwards will send me notes and say, thank you for saying that. Like we could all sense that this thing wasn't going the right direction and like we're fired up. We're on it. Like, and we're, we just redoubled our effort. Right. Yeah. People respond to emotion. Like if, if you're in the meeting, you're the leader of the company and like you're actually angry and like maybe even pounding the table or you're just like, it's like, oh, this is like, he cares about this. It's like very serious to him. The way I think about this, we talked about, I think I saw you in New York, uh, like two months ago and you're like, dude, I listened to your other podcast founders in the gym. And then me and you had dinner two weeks ago and you said, you're like, I'm like, I listen to founders in the gym. There is like a, like a hype. It's almost like there's like, I'm transferring the emotion and the excitement that I have from this crazy life story of this entrepreneur to the listener. Um, and one way I think about this that I think is really important, it's like the people I admire that the way they talk about their products, like Enzo Ferrari or Steve Jobs, I think I said in the Enzo Ferrari episode, it's like he's describing the products the way you talk about your lover. Like this is not a fucking car to him. In my own work, I'm like, listen, I want to be like embarrassingly passionate about what I'm doing. Yeah. I think that's like the mark. It's like, am I embarrassing and passionate? We were talking right before we start recording. It's like, these aren't interviews. These are conversations. And it was like, well, how do you pick who's on the show? People, it's just like simple, uh, like simple question, who am I intensely interested in talking to? Yeah. That's it. I don't know if you feel this way about your work, but I kind of always feel like it's not good enough. Right. Like I'm constantly like stressing about things in the Coinbase app, like, oh, this thing sucks. It's too slow. It's broken. Why doesn't this feature work? And I'm just kind of obsessive. It stresses me out. But I use that as motivation and feel right. And I'm like pinging, you know, 30 or 40 people a day about like, how's this part moving along? Why isn't this part, you know, and trying to create that sense of urgency in the company. But I don't think I'll ever feel like it's done or good enough. Like to me, it's, it's so, it's still so broken based on what it could be, you know, all the top performers are like this, uh, David Ogilvy has a great line about describing these people that are just creating some of the greatest art, greatest products, greatest ads, greatest companies. Like they're divinely discontent. Yeah. And I was just on the phone. My friend Kareem Atiyah is, uh, uh, as the co-founder and co-CEO of ramp. And he's like this, like we, every time in Miami, we spend a bunch of time together. And to the point where actually, uh, I guess I'll talk, tell the story, but I accidentally made a bunch of money because, uh, I was at Kareem's house for dinner and it's always like, it doesn't matter how good, how fast the company's growing and they're creating all these new products. It's just like, everything's fucking broken to him. He's like, doesn't like anything going on in his company, which again, that's very common because you see everything that, you know, you see the real thing, but then he also like has a very high bar. He's like, doesn't like other products, doesn't like other founders. And I remember hitting the table at dinner. I was like, God damn it. Like, just tell me who you actually like. Cause I'm like, if you hate everything, if you tell me like something you like. And he told me about, uh, this guy named, this founder named James who did this company called profound. And so I, I pull up X and I was like, I got to see if Kareem is saying this guy's good and they use the product, ramp uses the product and, and he likes it. That's a, that's a good signal. And I, uh, I see the guy had already followed my founder's podcast account. So I immediately follow, follow him back and then he hit me at the DM. He's like, dude, I've been listening to your podcast forever. And he's like, we're about to close around. Do you want to invest? So Kareem's a divine discontent accidentally made me a bunch of money. It's one thing to be perpetually dissatisfied and never satisfied, but not like devour essentially like turn that almost into like this negative inner monologue. So what's it like for you? Like, okay, I'm pissed about all these things that need to be better in coin, in the coin based app. But like, how is it internalist is like, are you super negative for yourself for letting that happen? Or it's just like, or it's just this more like motivational to like improve. What is my inner self-talk? I mean, sometimes I'm choosing consciously what my self-talk is, right? Like in the morning when I wake up, I'm trying to like get my head right for the day and do some stretching, do some exercise, listen to some, you know, listen to founders, listen to audio books and jumping in and starting to do some calls and work. I'm basically like repeating things that I want myself to believe and feel even if something doesn't go well. I'm not like beating myself up about it, but I'm just like, I just like, fuck, we got to make this better. And like, let's, let's get after it. And I'm always thinking like, what's next? What's next? Even when something goes super well, I often don't really take a moment to like celebrate very well. I'm just like, cool. That was great. What's next? What's next? You know? So I don't know. Somebody could say, well, that's not being, you know, in the Buddhist sense, not being present in the moment and appreciate. But I also think it's fulfilling over time to try to go do things that make, you know, you want to go do. And like, I do feel a deep sense of fulfillment over time at any given moment. It can feel like, man, I got my ass kicked today. Or like, that was pretty good. Like a bunch of cool stuff happened. But over the arc, I think it's like the goal to me is fulfillment. It's not like happiness or something. And you're not going to be getting fulfillment out of doing something that's, that's easy, right? If I was just chilling all day, I don't think I'd feel very fulfilled by that. Well, tell me the difference between fulfillment and happiness, though. I think of happiness as kind of like a fleeting temporary high or something like that. But fulfillment is like a deep satisfaction that can come from, I don't know, doing, doing meaningful work, having meaningful relationships, seeing things progress over the long arc, being useful in the world. Like, that's probably one of my, my values, right, is like, I think everybody should try to contribute to civilizational progress and, you know, help our fellow people that there's a sense of fulfillment that comes from that. And that's a high bar, by the way. I mean, you can do it in small ways, like you can just like, I don't know, cheer up somebody by saying thank you. Or you could try to make a new financial system for the world that's more fair and free and efficient or whatever, you know. Which is your mission. Yeah. And, and try it. But, you know, I think we're like 1% of the way there. It's like a massive thing, right? So, the point of the game of life is not happiness, and it's not comfort or anything like that. It's, it actually should be difficult. It's like a video game. If it was like easy, it'd be boring, right? So, you want to lean into things that are outside your comfort zone continually to try to make progress. And that's a sense of fulfillment. Yeah, there's a great story that Daniel Ack, founder of Spotify, told on the very first episode I ever did of this show, where Dara was offered the job, the CEO of Uber. And I think originally he said, you know, I'm comfortable to Expedia or wherever he was. And him and Daniel were having drinks and Dara tells a story where it's like, Daniel put his drink down and looked at him. And he's like, when is life about happiness? It's about impact. He's like, this is one of the most important companies in the world and you can have an impact on it. And so, his whole thing is like, yeah, you should optimize for impact over happiness. One of the things I love about Ramp is their mission is easy to understand. They want to help your business save time and money. And that is why they have built Router. Do me a favor, while you're listening to this, go to router.com. Router was built to reduce your company's inference costs by matching every request to the most efficient AI model. Router automatically balances cost, speed, and quality of output. Router is built for CTOs and loved by CFOs. Your engineering team gets the best model for every workload and your finance team gets lower inference spend. One customer of Router has the perfect description of why your company should be using Router. Choosing the right model makes a meaningful difference to our AI spend. We run billions of tokens through Router and have reduced our model costs by 92%. Tokens are money. Save both. And you can do that by going to router.com. Deal is how the best founders turn the world into their talent pool. I've been studying how history's greatest founders operate for a decade. And one thing they all have in common is they understand that recruiting and hiring the very best talent is your most important priority. A players recognize other A players, which is why top companies like Ramp, Shopify, Eleven Labs, Uber, and DoorDash all use Deal. Many of the top founders I know have personally invested in Deal after using their product. And what they discovered is that Deal is the best company in the world at building infrastructure for global hiring. Deal will help your business hire, pay, and manage any worker anywhere in the world so you can retain the best talent anywhere and spend the rest of your time focusing on what you do best, delivering value to your customers. The founder of Eleven Labs has a great description of the value Deal can give your company. He said, we built Eleven Labs to break down language and communication barriers. With Deal enabling us to hire and support exceptional talent anywhere, we can accelerate our innovation and bring more voices, stories, and ideas to every corner of the world. Deal is trusted by over 40,000 companies and growing fast. Learn how they can help your business by going to deal.com forward slash senra. That is deal.com forward slash senra. Going back to when you're upset at the quality of the product or you know, you see all the flaws and things can be improved, like, take me through like how you start thinking about that. You said it's not a negative monologue, but it's just like, are you excited because you see room for improvement? Like, what is the actual like response that's happening inside of you? Some of it is like a competitive thing, I would say, with other products I think that are out there or like how we make it better. Some of it is just even if there was no competitor, I think I'd still be pretty motivated to do it for some reason, because my brain is like very obsessed with building things in progress. It's like growing up playing video games like Civilization and StarCraft. And it's like, it's basically just you collect resources, you build industry, you expand to the next thing. It's addictive to my brain. Like, I just can't turn it off. I don't really play those games now because they're like, I don't have enough time. But there are like fun games like Polytopia, which Coinbase sponsored the world championship of, which was very fun. You know, I also don't have time to play, but these things are fun to just like for my brain. And so I think of those games are, they're scratching the same itch, but they're very short term. It's like a 20 minute, 20 minute game or something. But in business, it's very much the same thing, but you do it over the arc of decades or hopefully a lifetime. So what did you learn from these video games that you applied to building Coinbase? Civilization is a very interesting game. I don't know how many of your listeners have played it, but there's a tech tree, right? And you're like, I actually have a printout of the civilization tech tree in my office at home, because I think it's such a fascinating concept where you start like civilization started with like agriculture and writing and pottery. And at the end state, you know, you're doing like AI and biotech and like, you know, financial markets and nanoscale manufacturing or even going out into future tech that hasn't been invented yet. And so in some ways, like playing that game growing up, I was like, I want to add a leaf to the to the tech tree and like expand it a little bit. Like that's how civil, like, I don't know, civilizational progress is just so, so like fascinating and exciting to me. And the other fun thing about that game is there's various ways to win it, right? It's not so zero sum, like you can have a military victory and conquer, but you can also achieve like a scientific victory. You can achieve a religious victory. There's various ways, diplomatic victory. So there's different ways to win it. And I'd say strategically, I'll give you like one example. I remember this from a particular game of civilization I played as a kid where you're on one continent and you're battling with another civ on that continent and you're sort of consuming all of your resources to be on that one frontier. Meanwhile, you notice on another continent, one of the civs is starting to take over. and it's going to become massive while you're engaged in this other battle. You have some technologies you've achieved on the tech tree. You can gift it to the weaker opponent on the other continent so that that battle continues and occupy your future opponent so they can't get too out of out of hand and large while you're finishing this battle over here. Right. So there's just like there's so many layers to the game. I find it fun, but you can learn some simple strategies from it. Do you feel with the proliferation of A.I. it's making, you know, building and running a business more video game like? Kind of. Yeah. I mean, you can do more of it solo. Like there probably are going to be like these companies like one person billion dollar company. I don't know if that's actually technically been achieved yet. If not, we're probably close. But yeah, I mean, the thing that I starting off as a programmer, I was doing lots of solo projects and just, you know, it's good for introvert mode. But I realized at some point to achieve more and advance in the game of civilization or business or whatever, you know, I had to find teams of people and work with them and empower them. And that is its own whole skill set that can be very frustrating for a nerdy introvert that's like, oh, why do I have to communicate through this channel and get this, you know, and then if I had just done it myself, it'd be faster. That's the kind of initial thing a lot of managers struggle with. So, you know, I really have made peace with that. And Coinbase is what are 4000 people now and we can always make it more efficient. We should talk about how is doing that. You can also have a team of agents, right, that are working for you 24 seven. And actually, like many employees, including myself, have that now Coinbase, which is its own fascinating thing. Actually, let's talk about that now. So when we were hanging out a few weeks ago, I was like, dude, I want to do an entire episode because I've just been loving your tweets. And I was like, this is going to sound weird for an episode. But what if I just read your tweets back to you and then you expand on them? So the first one, this one went viral. It got like a million or two million views. And it starts out with I've changed how how I work over the last few weeks. So I'm going to read like just a paragraph or two. It goes on for a little bit. But the ending is really interesting. You said what an exciting time to push the frontier of productivity, especially as we build the economy for agents, agents and finance. We should all keep pushing it direction. So that's where you end. But you were talking about like, hey, I have one lead agent for each area of focus running all the time and a chief of staff agent that oversees them. Each lead manages its own sub agents that write code, test it, chase reviews and coordinate with team members. This goes on and will link in the show notes to your tweets. But can you go into, hey, I've changed how I work over the last few weeks because of AI. When we first started using AI at Coinbase a few years ago, the normal way people use it is you prompt it and it gives you a response. And what we did was we eventually shifted this to where you give the agent an area of responsibility or a goal, and it can have a team of people or agents under it. And it actually works on that area. And periodically, it'll come back to you with important decisions like, hey, I know you want me to optimize onboarding at Coinbase, so it works better in all these countries around the world. I looked at the data. I analyzed it. The five biggest opportunities are these. What I'm proposing to do is start with numbers one, two and three. Do you agree? I'm like, yep. And I click and I click yes. And so it is coming back to me with ideas about what to to work on. And I and I aligned it. I got I guiding. I'm guiding it. Right. Let me pause there. Yeah. So the quality of the suggestions, do you put it on like C player, B player? Like right now, what you're getting, you're like, goddammit, this is like a genius level employee already. When you're doing the planning, right, you're typically picking like one of the best frontier models that's the most expensive. It's putting together. It's investigating, coming back with higher level ideas and tasks. Now, as it gets subtasks, it can now farm that out to the lower level models. Most employees are using what we have is like auto in there. It's choosing the model for you at this point, because I think even the overhead of a human being having to choose which model for which task is like an unnecessary task. At this point, I should mention one other thing, which is we're using a lot of open weight models, too, and actually fine tuning them to do specific tasks trained on internal data. And in some cases now we've seen them outperform the frontier models like a much smaller open weight model that all it can do is file and draft compliance reports at Coinbase. That's all it knows how to do. But it outperforms the frontier models because it has been trained on tons of internal data. So we're getting teams of agents like that, too. And those open weight models tend to be much faster and much cheaper. But before we go to what the employees don't have, like what you're using, you're set up going back to the tweets, like I've changed the way it worked. So like when you're getting these responses back, hey, Brian, this is what you want me to do. This is what I think we should do. The feeling of it is like, holy, are you like shocked or you're just like, OK, this is like a normal, you know, regular employee. Well, it quickly becomes very normal. I mean, yeah, it's like working with other people. But like the quality of people is what I'm trying to get to inside companies. A lot of companies in tech use these levels like in internal come in at level three or something and a VP will be like level 11. I think these are easily like L5, L6 employees. Yeah, I think you can get almost like an infinite team of these L5 type employees now. OK, and you think that's where like this high as it goes right now? You haven't had an experience was like, damn, this can be like my right hand person or somebody like a direct report running a big part of Coinbase. AI is not exactly like a human. It's kind of like a submarine is not exactly like a fish, right? It's like it's it's going to be it. In some ways, it's far exceeds human performance already in the sense that it well, it works 24 seven, for instance. It gives me answers very quickly every time and it can make infinite copies of it. Right. Or spin up as many of them. So in that sense, it already exceeds human performance. In other ways, it's not there yet. There are definitely times where it misses the big picture, right? Or it feels a little derivative or something of other stuff that's out there. The pace of improvement is very impressive. So a lot of people using the generic models that come off the shelf, they've been trained on sort of the corpus of all the data out there in the world that they can get their hands on. But they're not necessarily like tailored to your preferences and your company's way of working. So what we've been doing, in addition to fine tuning our own open weight models for narrow tasks, and by the way, we'll get back to in a second how I think this is going to create a new agentic economy, which is important. But we also are developing this thing called like Coinbase Brain, which is a service internally that's getting context. And there's a brain now for every employee, every team and every GitHub repository. The idea is that as humans are working on these teams or these GitHub repositories, the brain is accumulating and distilling lessons and knowledge. There was an A-B test we ran. There was a user study we ran. There was an incident that happened, an outage. There was a meeting and like a meeting transcript. Like these things can all be ingested. Code reviews, you know, these kind of things, even like policy documents that come, you know, like gap accounting standards or whatever. So it can kind of ingest all this material. And every night it sort of pours through it and distills down key lessons and amends it into the brain. Then when the agent goes to work on something in that area, it puts the brains into its context and it creates the output that is more aligned to what we as the company internally are trying to create. And so our jobs increasingly as human beings are to align the A.I. by editing and cultivating what's in the brains and resolving discrepancies. Sometimes the brain has a contradiction in it. It's like, well, you know, David said this and Brian said this, like, you know, 14 months ago, like, which one do you want to do? You know, these kind of things. And it's basically improves the rate at which the agents can one shot the work without a ton of like feedback from the human manager. Now, I should clarify here, like the humans are making the final decisions. The humans are making final code reviews. So it's eliminating tasks and it's getting to the answer more quickly with with less kind of nudging and corrections. But the humans are still very much the ones deciding the direction. And that's our main job now is, I think, to like align the A.I.'s that are working on the software factory that produces the output. How do you think about what the brain is? It's just the operating system for the company now? Yeah, I mean, you could argue that it is like what is the value of a company? It's like it's a set of practices and people and know how that have accumulated over time. And in some ways, it's formula. It's like formalizing that, codifying it. But yeah, I mean, you still need like people to actually go implement the thing or agents in these cases. So I don't think I don't think it's ever going to be like done or fully replaced. Like people have this question a lot about could you run Coinbase with 100 people or something? And I don't know. I mean, my guess is we'll probably keep roughly the same amount of people, but they'll all just be able to do way much way more. Yeah, we'll see. I found one of my all time favorite quotes when I was reading the book, zero to one, the quote says the single most powerful pattern I have noticed is that successful people find value in unexpected places. And they do this by thinking about business from first principles instead of formulas. That is exactly what Apple has done with their advertising platform. Apple oven connects you with over a billion potential new customers in mobile games. Apple oven allows you to capture undivided attention. Apple oven ads are full screen videos that are watched for an average of thirty five seconds. That is retention that blows other ad platforms out of the water. And you can launch on Apple oven in minutes. You set the goal and Apple oven achieves it. No complex setup, no expertise needed. And Apple oven scales quickly. They can put your ads in front of over a billion potential customers. Other businesses have seen immediate results scale to hundreds of thousands of dollars of spend per day and increase their revenue by millions. So you want to get started quickly before all of your competitors are on Apple oven. And you can do that by going to Apple oven dot com. That's Apple oven dot com. So go back to say you wanted to get into the fact that this is going to enable this agenda economy that you think is important as you start to get more and more agents that are working with you. By the way, the UI has to change for that, too, because this quickly overwhelmed me once I had like more and more these chat sessions open edition. It was a lot of context, which I'm just going through all these walls of text to see what was going on. So could you take me into like how many how many of these things you have open at one time, like what's going on? Yeah, well, right now I have probably seven or eight lead lead agents and then that are working on different areas that I'm focused on right now. And then one chief of staff, which kind of oversees them. But each of the leads has they can they can spin up agents to get subtasks. So I don't know. It depends if all of them are running. But anytime when one of them is running, they maybe have like three to five different sub agents or something. So it quickly got to a place where I just couldn't keep all the all these walls of text, like what's going on. So we've built this kind of internal view that we're we're prototyping, where it's more like a queue or an inbox of things that you need to review and decide. And they're prioritized, too. So if you have like a big backlog of them, you only have five minutes to like you can knock out the decisions that would help move the team of agents forward the most. So, you know, it might it might say, here's here's a product review. Like, here's a couple screenshots of this feature. We just developed. And the next step is to do this. And I'm like, yes, no. Or I can do a little voice dictation to like give it a couple bits of feedback verbally, which is very fast, faster than typing. Right. And so you can say like, yes, no, no, no. Yes. Do this one a little bit differently. Just start knocking out decisions, almost like in your email inbox. Right. But using keyboard shortcuts and voice dictation is is I find to be faster. And then the agents sort of keep track of all these projects that are going on and decide who else on the team they should be collaborating with, getting approvals from. So that's reduced a lot of the cognitive overhead. Why does the UI have to change? Well, I think you can't just deal with all this wall of text. Yeah. So what does what like what's the UI that you want to see? Well, I think having a queued inbox that you process, it's a prioritized set of tasks or decisions with very simple. It's like everything just appears in the same screen at the same time. The buttons are in the exact same place. And you say one, two, three or, you know, that that alone is good. Now, I guess I have been thinking a little bit about like what's the next layer beyond that? Because. I mean, I think it's sort of like getting even higher level employees up that stack, right? I mean, you might be able to have like a VP level or something. There's just like, I think we need to expand into Brazil. Go do it. I'm going to check in with you every three months. Like that would be another layer of the stack, right? Where that person's like, all right, what's the team I need to recruit? Who are the partners I need to go meet with? Where do I need to sign these deals? You know, so and there are still some areas which are a little bit difficult to like prompt into existence. Like software is getting easier and easier to prompt into existence. But like regulatory licenses are not you can't prompt those into existence yet. Right. KYC funded accounts like we have to think about what the defensible parts of the business are in this world, right? Like the liquidity, the network effects from having like the most liquid exchange in the U.S. or USDC has a big network effect, like things like that. Right. That interface has gotten more and more. Important, who's building the best interface for this? I mean, a lot of folks are trying it right. You've seen like Grokbot kind of, I think, pioneered this like chief of staff idea with all the different folks. I haven't seen somebody do the prioritized inbox kind of queue thing yet, but that probably I'm sure there's other people playing with these ideas. We're working on a little prototype, too, of like almost like an operating system for the company that all employees could jump in and view. And it's really just like, OK, what are our top priorities? What are all the projects that are contributing to that? You can drill down and say, what's the what are the sub teams, what the individual people, what are each of the tasks that they're working on? Who needs to sign off on what? What's the most important thing at any given moment? You know, or what is the critical path like from here to there? What is the longest set of dependencies? Who's working on them? The AI could even come back and say, here's the biggest risk in the company. It's like this is a critical dependency on 14 downstream things. And right now it's understaffed and its velocity is below the average of the company. And it's like proposed solution. Move, David, over to that team. Number two, change out the leader. Number three, change the strategy. Number four, narrow the scope. And you're like and it's like, I reckon number three. And you're like, yes, and it just does it. So you can get to a world where you could be like, you know, like Steve Jobs is like bicycle for the mind type metaphor. It's like this is allowing people to have much, much greater productivity. As the models get better, they'll start to be that that VP level. I was going to say earlier, you know, on coordination amongst agents. I think if you think about this is all happening within one company right now. But agents are also starting to interact with other companies. And so you can kind of think about the metaphor of like there's like an individual person working by theirself on something. Now you can get to a small team of people. You can get more done than the one person. Hopefully, you know, at a company, you can get more done than a team. A country could get more done than just an individual company. Hopefully the whole world, collectively humanity. We can actually get a lot done even more than one country because lots of dependencies between countries like trade and. And so, yeah, I mean, how do we start? to get an economy or teams of agents working efficiently. We're seeing it now within one company, but as they start to interface with other countries or companies, they're gonna need their own markets and financial infrastructure. Like that's actually, how do you get 8 billion humans on the planet to all, in some sense, like collaboratively create an economy for the world? They need markets, they need money, they need specialists, right? Who can like come together, like a great security person, a great designer, a great financial analyst, a great product person, and then come together on a team to achieve more than the sum of the parts. So that's kind of what we're doing now with building agentic finance and the payment rails for that, which I think credit cards are playing a role early on where it's like, hey, buy me these pair of shoes or book me this plane ticket. Credit cards have a lot of importance for backward compatibility, but it's also gonna be like stable coin rails on crypto. And I think that that's gonna be like the real time payment rail for the agentic economy. But if you can spin up an infinite amount of agents and you actually have a functioning agentic economy, wouldn't that economy, like it would dwarf, it'd be much larger than our existing human-based economy, right, or no? Yeah, I mean, someone was asking me today, like what percent of transactions in global GDP do you think will be agent to agent in 10 years? And I was like, five nines, you know, 99.999 or whatever. So I think that tipping point could happen sooner than we think. But yeah, what was the follow-on question? Like, you're wondering what's the implication of that, or yeah, what? No, just like, you spend all day, I think another trader's gonna ask you about it. You're like, I wake up every morning thinking about agentic finance. I wake up every morning thinking about podcasts. Yeah. Like a little different. Both doing what we should be doing. Yeah, so I'm just like, okay, well, like, if I'm trying to process like what you're saying, where you're like, okay, well, every agent could spin up an infinite amount of agents and copies as much as it wants. And then if you do have one where it's kind of, you know, it's free functioning, you don't need a human in the loop, like, well, then we can make way more transactions. It's like, the economy would be bigger if instead of 8 billion people, there was a trillion people. So like, why wouldn't that be the case if you can actually build what you're talking about building? Yeah, I think that's exactly right. And that's why I'm so excited about it. And I think crypto and AI really are like, better together in this world. You could say that what Coinbase is doing now is we're taking frontier technologies like crypto and AI, using that to build better financial services. And ultimately, in service of creating more economic freedom in the world, right? Like, that's our mission. I think, in the same way that crypto like, decentralized the financial system, and I think improved freedom, I think that we have an obligation to kind of make sure AI is decentralized too. Because the bad version of it, I think would be, there's one or two big players, they're kind of like, have too much control, they could be biased, you know. In a decentralized world of AI, everyone has lots of small open weight models. You control it, you can customize it as much as you want. There's an economy of these agents all talking to each other, which means you have choice. You have choice in the economy. Do I want to go with FedEx or UPS? You know, you would choose like, different agents with different specialists, different providers, and assemble them using the financial markets to come together. A lot of folks right now, I think, are in the financial services space, they're a little like, skeptical or afraid of AI. And they're saying, we don't want those AIs to touch our accounts, because they might do something bad. Now, we are building very strict controls and permissions, and like, the customer always has to be in control. But to me, this is an opportunity. I'm like, we should lean in and be the most AI-enabled financial account, and make it safe and trusted. That's what Coinbase can do. That's what we did with crypto. I was gonna say, that's what you do with Bitcoin. Yeah, that's what we're sort of known, we're known as the most trusted crypto brand, and we have, we store more crypto than any other company. So, why can't we do with AI? I wanna go back to this idea that you said earlier, what if you had a VP-level intelligence for AI, and you said, we should expand to Brazil. Just go do it. Have you spent any time with Scott Wu from Cognition? Little bit, yeah. So, I'm gonna see, he's coming back on the show in like, two weeks. I think he said it on the show. Maybe he said it off-camera, but he's like, it's just a matter of time, where it's like, you're gonna have an agent where you set it on a task, and it just runs on that task for an entire year. Like, it could build a company. It could be doing a lot. Like, what could a human get done, a very competent human get done in a year? I had a weird thing that came to mind. I'm listening to this audio book where the President Polk, I think, I don't know, he's like the seventh president of the United States. He's doing the Westward Expansion. He really wants California. Mexico owns it at the time. He sends like, a hundred soldiers out there. Just figure it out. Go from Washington. Go take California. And then it takes these guys like, two years, but the hundred people wind up taking California. They lose it, but I thought like, oh, that's kind of like, there's an analogy here. It's just like, okay, here's a hundred really competent people. I don't, I won't know what you're doing for two years. You kind of come back and they actually have a new state. Yeah. Yeah, high agency. I mean, that's like one of the most important things to look for in anybody that you work with. And I guess in agents, that's what the word agent, it has a, it's a model with agency. What is the thing that you've had an agent do inside Coinbase that most surprised you? Okay, well, I don't want to say the name of the vendor. It's probably bad for them, but we have so many vendors and like, vendors have their own cybersecurity issues. So I like the idea of having fewer vendors and having control of our own software. And if we can save money on it and have better security, like, let's do that more. Say more about limiting the amount of vendors. There's been a whole massive increase in cybersecurity events, right? We just had Nikesh Arora on the podcast. Yeah. He was talking about this. Yeah, yeah. I think the number of cyber incidents, both in crypto and tradfy, by the way, it's happening everywhere. It's like roughly doubled year over year. October, in the industry, they call it like black October, because it was like every week there was multiple. Obviously, there's been a few in the crypto space too recently. And security is our kind of a bread and butter as most trusted brand. And we use AI agents through like frontier agents to review every line of code before it goes out in production to look for vulnerabilities. We store 99% of the crypto entirely offline, disconnected from the internet, that which like air gapping it distributed, decentralized around the world, resistant to natural disaster, nuclear war, all these kinds of things. And there's a lot that we've been investing in security for a long time. We realized at a certain point, one of our biggest risks was actually third party vendors, because we have less control over that. So the vendors we work with, and generally we hold them to the pretty high standard, they have to compass these kind of security audits. But we're looking for ways to just reduce the number of those vendors by rebuilding it in house, because why take the risk? It seems like it's also more fun for you. Like now, like it seems like you're having more fun because you're just building more shit. Yeah, I mean, the cyber part of it is that scary, but we got to be vigilant about it. But yeah, in general, building, I'm having lots of fun. It's been very energizing. I almost worry about it sometimes, because like I, yeah, I mean, I'm just so excited. I can't sleep at night. I'm like, this is so interesting. And then, okay, I have to go to sleep. I have to go to sleep, and then I wake up and the first thing I think about, it's like, yeah, we're living in a golden age of being able to build in productivity, innovation. I really feel very lucky to be kind of, I don't know, hopefully in the prime of my career, kind of in this moment, because it's really cool. I spent the last decade reading more than 400 biographies of history's greatest founders, and if I had to distill what I learned down to a single word, it would be focus. The best founders spend all their time on the thing that makes them great. They spend their time building products, serving customers, and growing their companies. The greatest founders would not spend time rebuilding authentication infrastructure. That is why more than 3,000 of the fastest moving and most important companies in the world use WorkOS. Companies like Anthropic, OpenAI, and Cursor all use WorkOS. WorkOS makes it easy for you to sell your products to enterprise customers. If you build software and you want to sell to the largest companies in the world, there are some requirements that you cannot ignore, like single sign-on, SCIM, role-based access, audit logs, and more. You could spend months building all this yourself, or you can use WorkOS. WorkOS provides the infrastructure to make your software enterprise-ready. Go to WorkOS.com and get started today. That is WorkOS.com. This is something I asked Sam Altman when he was on the podcast, where it's just like, everybody uses AI, everybody hates AI, right? It's like, obviously you got this. I don't hate AI. No, no, no, no, no, but like, this is an audience full of entrepreneurs. Like, if entrepreneurs aren't using AI, they're fucked. That's, I mean, the general population. Like, you're super excited you can't sleep, right? Because these are tools, and they drastically increase to how creative you can be, and the stuff you can build, and the problems you can solve. Like, why do you think there's such a disconnect between the way you and I feel about it, compared to like, the average person? Most people are wired to be a bit cautious about change, right? It's probably like, you know, I find like evolutionary psychology to be very interesting, right? A lot of explanation of human behavior can be traced back to evolution. So, probably people who were somewhat fearful of change lived longer, and reproduced more, and there were more of them, right? Whether it's like a giant flood, or half your tribe disappeared, or, you know, like, these things would all be scary and bad in like an evolutionary psychology sense. So, luckily there are a small percentage of people in society who are, for whatever role of the dice, you know, genetically, or psychologically, or whatever, kind of are excited enough about it, because society needs some of that to advance, and you don't want everyone to be like a crazy entrepreneur, you know, that might be totally chaotic. That'd be fun. It'd be good for your podcast. Yeah, yeah. Eight billion people listen to Founders now. Yeah, yeah, imagine that. The reason I brought that up with Sam Altman is because he said something on the podcast, like, there's like a billion people using Chet Chibuti every week, or something. It's like a crazy fucking number. Like, it's a big number, and I'm like, but everybody's using AI, and everybody hates it. Like, this, outside of entrepreneurs, and you know, creative people, and stuff like that. It's a very interesting disconnect to me. Totally. I mean, it doesn't also help when they fucking go around telling, they're gonna take all your jobs and kill you. But that aside, there's also, like, every single person that hates AI, or the people that do, can't have all heard that message. It's just confusing to me. Obviously, I do think their messaging could have been better, but there's something deeper going on. Even if they had perfect messaging, I think people would still feel some fear and resentment about it. I remember talking to one of the board members of Amazon one time, and they were saying, you go into a room and say, you know, how many people are worried about Amazon, and, you know, local businesses, or like, you know, the workers, and like, a bunch of people would raise their hand. And then they say, well, how many of you used Amazon the last month? Everybody in the room was like, okay. So, people can hold these two things that are in their head at the same time, which is like, this is a useful thing, and I have some concerns about it. I guess it's that simple. There's like a great ex-account, I think it's called the Pessimists Archive, or something like that. Yeah, yeah. You can go back and look at things historically, when the bicycle, I think, came out in America, people were like, freaked out about it. And they're like, what, bicycles? And it was like, all these fears about, you know, you won't know where your children are anymore, they could be anywhere in town, licentious behavior is going rampant. Marc Adjuster's on the show, and you know, Marc, you know, because you've been on the show now twice, like, we don't know what we're going to talk about, like, I don't see any fucking questions. And so, I was just prompting, I was like, I know there's a lot of business history in this guy's head, and that's like, obviously, I'm embarrassingly passionate about business history, so I was just like, prompting him, he's like, do you know about bicycle face? And it's like, it was literally in newspapers, where they're like, you're going, you know, before you were walking, or maybe you're on a, like, you're going so fast on the bicycle, that the wind, and like, when you're like, you know, going fast, spreads out your face, it's going to get stuck like that, people legitimately thought that. And then the other story he said, was guys did not like it, because, you know, all the women are in the village, it's hard to get from village to village, now you have this girl on a bicycle, she's going to the village to see these other guys. It was just a hilarious story. You could imagine someone at that time saying that, this was going to disrupt the fabric of society, right? Or, I remember when I was a kid, like violent video games were like a big fear thing, and my mom was like, scared to death, of like, I was playing Doom on the computer, and these things. There was congressional hearings about this. Congressional hearings. When cryptography, like first came out, you can go back and read the articles, where Congress almost banned it for commercial, like, or civilian use, because they're like, this is going to be used by terrorists, and, you know, like, well, I just don't want someone like to read my email at the coffee shop, like, can we, you know? So it's amazing how good we are as humans of like, just scaring the hell out of ourselves about the unknowns, you know? And then, you know, the precautionary principle is like, well, we should enact all of these regulations, because we, in case something bad happens. But then oftentimes, like, the thing you're imagining in your head is not actually the real risk. It's either something totally different, or it's something much smaller. And you can accidentally just kill whole industries. Like, the Nuclear Regulatory Commission got created to help us against, I guess, like, the next Chernobyl, but like, no new nuclear plants got, I have sort of like a fun list of these on my personal website. So I think the precautionary principle is super dangerous. You should wait to see what happens before you try to put a bunch of rules in place to prevent it, because it might be something totally different than what you think. Like, we're all pretty bad at predicting the future. And I mean, somebody would say, well, Brian, like, you know, what if it's so bad that, like, the whole world ends or something? And I'm just like, there's no historical precedent for that. I'm just kind of skeptical. And humans are usually good at solving problems, like when they're presented, you know, especially if they're like some kind of existential threat. So, you know, DHH? Yeah. He was telling me, because him and Toby, like, they travel together, they race cars together, they're like working on, how do I say it, a machi? Yeah, I don't know why, sometimes people in Linux or open source projects, they do this, it's not how it's pronounced, and you have to pronounce it differently. It's just like, I don't know if it's a troll or whatever, you'd have to ask him, but I guess it's pronounced oh-machi. So they'll, like, work on it in between, like, when they're taking turns racing in the middle of a race and stuff, which I think is funny. But he was saying that he wishes, he's trying to convince Toby, because the internal AI tools that Shopify, they're building Shopify are so good, he's just like, this is better than anything on the market, give it to me, I'll buy it from you. And I think they're saying no. And then right before I came here, I was on the phone with my friend Kareem from RAMP, and they've now, they've been building a ton of AI tools, and all centered around, like, a problem they had inside their business that needs this, their organizing principle's very simple, it fits on a post-it note, it's like, does this save a business time or money? And ideally, time and money. And they decided, okay, these tools are so valuable to us, we're actually going to externalize them to the world and sell them, which area are you on? You're going to keep the tools yourselves, you're going to turn them into existing products for customers, like, what do you, how do you think about this? We had the same debate internally on a couple of these projects, like that internal harness I just told you about with the prioritized queue and stuff. And where we are on it is that, I, it's not that like, um, I do think it could be useful to other companies, but it's actually like much more difficult to build something, a tool that could be used in any company versus what a tool adjust for our own company. Like it's not just twice as hard. It's probably like 10 X or 50 X as much work to like generalize it and make it good for and then deal with every company's security needs and different billing preferences. And so it's, it's like a non-trivial amount of work. Now the other thing we could do is just, just open source it or something. We could do that, but then you start to get tons of people making, you know, requests of different. It is, it is additional work. I also think that a lot of these like coding harnesses and AI tools, they'll end up being open source versions of it, which means every company should really just run that internally. It's not, I don't think that like that's going to be a massive revenue driver or will look a huge part of the value chain might not get captured with these coding harnesses. So that's just one example that we looked at. There are other examples where we built certain infrastructure, like we had just to get our own products working. We had to connect to like every single blockchain and make secure custody and make staking infrastructure. We've now exposed all of these through our, our infrastructure product called Coinbase developer platform. So there are lots of companies building on that. We've got like seven of the largest banks in the world now building on top of that for like stable coin wallets and payments and trading things. Is that going to turn into a big business for you? It's still pretty early, but it's growing really well. Yeah, I think that's going to be, it's going to be kind of like AWS for us. I'm pretty bullish on that product. So a lot of the crypto infrastructure we've built, we haven't really put any like of our agentic tools yet, but we'll see maybe. Okay. So still undecided. Yeah. Okay. I haven't seen one yet that it shouldn't just be an open source free project. The reason I asked you is like, how do you think about Brain? Is that like an operating system? Because he went into detail about like how he's running the company and they were like building AI tools in like 2011. And he essentially has inside of bending spoons, like proprietary AI operating system and it's getting better and better. And so he feels like that's why he's able to do the acquisitions he's doing and then turn it in for the product and make them way more profitable and everything. And I asked him, I was like, would you ever sell this anywhere else? Like it's not valuable anywhere else. It's like so fine tuned to us and us using this as opposed to selling is like a distraction because I should just buy the whole company and then layer this on top of it. Yeah. Very similar to like Alex Taubman of Long Lake, who was on the show. Same thing. Like they have their own internal AI operating systems and he found it better. Like if you want to induce an organizational change, it's a lot easier to just own the company you want to change as opposed to selling the tool to other companies and not having control over how they use it. Yeah. I thought that was interesting. Yeah, totally. You have any thoughts on this? When selling B2B software, you will get a very long list of customization requests, you know, and I feel like it's taking all of our time and energy just to stay on the frontier of our own stuff. And the tool is better for you because it's hyper specialized to what you you're doing. It knows everything about your company and what you're trying to do in the world. And it can make our own products where we have already billions in revenue kind of much more efficient as opposed to trying to spin up a new product line. There's always a balance of how many things you should be doing. There's value in focus, but you can also get really powerful effects by having multiple pieces and being one of the only companies at scale, like economies of scale that does all these things. So at the moment, I'm trying to be more disciplined about not launching too many more new products. We've got to focus on the ones we've got. You said something fascinating when we had dinner a few weeks ago about the fact that if productivity is increasing, why is revenue lagging? Do you want to elaborate on this? Yeah. I mean, I think a lot of companies are trying to figure this out right now because we are seeing material, like very material increase in productivity because of AI. Because of AI. And it depends on which firm you're talking about. But I think for us, it has not maturely moved revenue yet. I think there's two theories on this. One is that it's a lagging indicator. And we're just now getting like a step function change and it'll be there in six months or something you'll start. It's just a lagging indicator. The other theory is that because, you know, we're getting higher productivity, but others are also getting higher productivity, that it's like a Red Queen race, like you have to keep running faster to stay in the same place. I don't think it's the latter. I mean, there are some other companies certainly embracing and like leaning into this in a big way in the fintech space. But I don't see it happening as much in TradFi, you know. So I think it's probably lagging indicator and we'll see it show up soon. That's my guess. But it is like sometimes it's a bit of a paradox that it hasn't moved needle more. Got it. All right. I want to go back to another tweet. We got flying cars. Now it's time we get a modern financial system. Let me Coinbase pill you for a minute. Financial services is a massive industry. All fintech companies combined capture about 4% of total global financial services revenues, while TradFi, traditional finance services companies account for the remaining 96%. This was something I was shocked at. I think it was David Velez from New Bank that said it. And then I asked Nick from Revolut, who's also on the show, where all of fintech is something like four or five, less than 10%. And everybody thinks, oh, fintechs are competing with each other. They're all kind of going to this like super app kind of idea. And David Velez, and then I think you're making the same point, he says, well, actually the huge market to tap into and to grab share from is TradFi. Totally. I give David a lot of credit for articulating that point well, and he's right. I mean, the big opportunity is taking more share from TradFi. It's also, think about the TAM from the agentic economy I just mentioned, right? That's kind of like a whole new TAM that nobody knows the size of, and that's more like blue ocean kind of. If we can just, at Coinbase, be good at using these frontier technologies, crypto, AI, update the financial system, crypto are the rails, AI is the interface. If we can do that, the expansion is enormous, right? It's less about any individual fintech that we're competing with. When I heard that the first time and I read your tweet, I was like, how can that even be true? Because it's like, I feel like there's been fintechs around forever. Yeah. Coinbase is how old now? 14 years. 14 years. I think Revolut, obviously, same thing. Newbanks around that as well. There's a ton of other people, it's just like, there's all these other, you know, applications inside of fintech. Why do you think it's so slow and why traditional finances essentially still has the entire market? As a kid, I was always like, sort of assuming, if I saw something better, I'd be like, well, the whole world is going to be using that in a few years, and it turns out that the world has a lot of inertia. Maybe it comes back to this psychology point we were making earlier about people fearing change, right? If there's something they've used for a long time, like a lot of people in America still pay rent with like a paper check or something, right? It seems bizarre, but once you learn how to do something, and, you know, frankly, when you reach a certain age, people, sometimes they stop learning new things, right? They're just creatures of habit at that point. They don't want to change. So it takes like a generational shift sometimes for these things to happen. Progress moves a lot slower than, you know, in my ideal world would be the case. There's a founder of a very big company that's going to come on the show soon, so I won't say who it is, but the point that he made was if you keep trying to sell tech startups, you're going to build a very small business, and because he sells to just Fortune 500, he's like, let me tell you, Fortune 500 companies, they have a problem. When they find a solution, they think it's done. So like he has almost like no churn on his product, because it's like they're not out there, like looking around at what somebody else is doing. I mentioned James from Profound earlier. Me, him, and Kareem just happened randomly like two weeks ago, three weeks ago in New York, had lunch. And he was saying like it's fucking crazy who like one day in his engineering department, everybody's using Codex. Next day, it's called Code. Next day, it's something else. It's just like the switching. It's just intense. Yeah. We see a lot internally, too. People are jumping. It's like there's a new announcement every few days. Yeah. So in the real economy, if you have a bank, if you've been with whatever, JP Morgan or whatever, it's like, why would I even change? It's like, yeah, you have to wait till these people die. Like what? Yeah. So in any industry, if you just make something like a little bit better, most people won't switch, actually. It's got to be something dramatically better or different. It's like, OK, why did people stop doing Google search and using, you know, an LLM, right? It was like, oh, because it was something dramatically, it wasn't just a slightly better Google search. I mean, there's a lot of people who wanted to use crypto early on. That's fine. That's how we got to get started. Now we've kind of brought together the everything exchange. OK, that's a good step. But what's going to get them to switch is like, I think the, you know, the billion plus people using AI agents now, they're going to want to start to benefit from that in their financial account. Right. They and if the legacy players aren't going to allow that to happen and be full featured through the agentic interface, to me, that is a reason to switch. And by the way, your agent might help you switch. How does the human benefit for agentic finance? It can just help you make more money. Right. Like that's a good product. Selling money is very. Yeah. So I'll give you an example. Right. If you go into trade in a traditional brokerage or something, you have to be like, all right, I want to buy one share of Meta or something, and you go in there to the Meta screen and you type in one share and, you know, you hit the button. That's not how people are going to trade. OK, so I have to know what asset and what type of order I want to do and how much and like that's the level one. Level two is like, OK, I want to buy Meta, but I want it to be 10 percent of my portfolio and like leg into that position over time, like maybe when it when it dips by the dip, you know, like help me get there over time. So that's like level two. Level three is I don't really know what asset I want to buy and and and when, but I have a thesis that the bottleneck for AI is going to be energy and data centers, not some thesis or another thesis that maybe like I think I think the looks maxing economy is going to be massive, like the longevity drugs or, you know, whatever. So you can type in like a thesis about the world and it'll say, all right, to get exposure to that, like looks maxing, you know, Eli Lilly and Nova Nordisk are the biggest GOP one manufacturers. And here's like some other instruments that might be useful in the supply chain or whatever. We put together this basket of assets we think you should buy at this time. But, you know, take profit and stop loss or whatever. You're starting to kind of work your way up the stack. Right. And I think that like going maybe to level four would be I don't really have a thesis about the world that I have super high conviction in, but I want to talk to my smart friend who's really good with money, you know, and that's where you would go into your agent or see like a community leaderboard, like of different ideas that other people are doing and say, oh, that's smart. I could I follow I follow this person, I think, is a really good trader, has a great track record. They just did this trade. I'll copy trade it. Right. If you just want to talk to your agent about like, what should I be thinking about in the world? It's like everybody now has a private banker in their pocket. Right. And that's the only like wealthy people had. So when I meet with my financial advisors, they come to me with ideas. I don't have to come. I don't have to come to them with like, I know I want to buy Metta. They're like, we have this whole team of people who came out with these things like we think you should make the following ones or whatever. Right. So come to me. Bring me the ideas like I don't want to come to you with ideas on what to invest. So those are how I think it's going to go up the stack and is going to help you in a truly enabled financial account. It'll make you more money. It'll save you money on taxes. It's all these kind of like tasks that are just a nuisance to like, I know my paycheck got put here, but I know it's not earning anything. So if I move it over here, I get 4% on my, you know, why doesn't it just do all that for you? Right. So that's what AI is going to help you do. Yeah. Mickey Mocha said on this podcast, again, that FinTech put a bank in your pocket and that AI will put a banker in your pocket, which is very similar to what you just said. Yeah. So in this scenario, right, you're talking about what's missing and hopefully coming in like the financial aspect. To zoom out, what is like the thing that you want AI to do for you that is not yet able to do for you? I think we need humanoid robots, right? Because it can't move things in the physical world yet. And it can't get on a plane and go represent me in some other country. It's still purely software. I think AI and robots, that's how we get like a truly massive labor force. Right now, it's kind of just knowledge work because it's purely software, digital. It can do legal work, you know, financial analysis. But once it can start to physically go anywhere in the world and move things like to build factories or mines or run bio labs, that'll be the next big thing. That's the next step, function change. I mean, like at New Limit, the biotech that I co-founded, we sort of been debating when's the moment to bring in humanoid robots to the lab to try to really test it. And we don't think it's yet because right now, a lot of these humanoid robot companies are testing it on like the barista use case of like, can it make you an espresso? It can't do barista yet. So once it can do barista, I think we'll start testing it in the bio lab. But we're not there yet. Okay, one more of your tweets that I want to read to you. And this is related. I don't think we ever close the loop on like the frustration that I think any entrepreneur feels. They're always upset, never satisfied, always can be better. So you have a tweet that kind of hints at this, like I'm spending a lot of my own time on Coinbase app quality right now. There are a lot of pieces that add up, instances where it's unclear why funds are on hold, making it simpler to go through ID checks, app startup times of speed, an error that doesn't tell you what went wrong, a button that leads to a dead end. This is kind of like your frustration we were mentioning earlier, fixes are rolling out over the next few releases. If something that frustrated you recently, reply and tell me, my agents and I are reading them. Yeah. So one of my tweets is called retail app quality. And that was one of the ideas that came back with was, let's put out a post. And we have all kinds of ways to gather feedback, you know, like support transcripts and errors in production. But it was like, I wanted to actually like show the company a little bit too, that I was leading on this, because I've been telling people internally, we need to raise it. So I want to show that I was contributing, throwing my hat in the ring, you know. So it was like, all right, let's put out an X post. I mostly drafted that. I think the agent reviewed it. And then we put it out. I said, go look at all the feedback, summarize, bring me back the top five issues and fire up an agent to work on it. And it put five workers on it. A couple of those are, I believe, almost live in the app now. How many of the things that need to be improved are coming from you compared to being found by AI? It's a collaboration, like is the best way to say it. It's like with a thought partner or a teammate or whatever. So I'm oftentimes I'm pushing it along and I'm like, hey, I'm still really worried about this. Can you figure out what's going on with this team? Like, why is this thing not happened yet? Help me figure out if there's something I can do to help. Other times it's coming to me. I was like, hey, Brian, I know you wanted me to work on app quality or like make sure that our Coinbase for Agents product has full coverage across the whole app. I've been digging through the data and like, here's the biggest things I think customers would benefit from. So it's coming back to me. If I push it on stuff, it's a collaboration. Awesome, thanks for doing this again, man. Always good to hang. Thank you. I hope you enjoyed this episode. Please remember to subscribe wherever you're listening and leave a review. And make sure you listen to my other podcast, Founders. For almost a decade, I've obsessively read over 400 biographies of history's greatest entrepreneurs searching for ideas that you can use in your work. Most of the guests you hear on this show first found me through Founders.