How to Build Superintelligence Inside Your Company

1. 0:00 – 0:39

   Intro

   1. GTGarry Tan0:00

      How do you build super intelligence inside a company?

   2. DHDiana Hu0:02

      Part of the key thing is not to just use AI as a copilot. This is the, the thing where you use it as the building layer for everything, and you need to start recording all the artifacts.

   3. JFJared Friedman0:14

      It's like a shared organizational brain. It's like the closest thing to us being able to, like, connect our brains.

   4. PKPete Koomen0:19

      If you frame this as a way for everyone in an organization to get better at what they do using the, like, collective skill and instinct of the people they work with, it's incredibly powerful.

   5. GTGarry Tan0:31

      [upbeat music]

2. 0:39 – 2:15

   YC's AI Stack

   1. GTGarry Tan0:39

      Today we have a real treat. Uh, we have a special guest, general partner at YC, our partner, Pete Koomen. He created Optimizely, which was one of the first and one of the best ways to do A/B testing for apps and websites. And since then, he has gone on to create all of our agent infrastructure at YC, so literally all of our harnesses and how we use AI internal to YC. Pete, welcome to the Lightcone.

   2. PKPete Koomen1:07

      Thanks, Garry.

   3. JFJared Friedman1:08

      For the last few years since ChatGPT, YC has been funding mainly AI companies, and we've been... we've gone through, like, many different, like, versions of advice for them about how to build AI-native companies that build, like, mainly AI products. And we've gone on a crazy journey with them learning all of this. I think a lot of people don't realize that internally YC is actually building and using a lot of the same stuff that we're helping our startups build and use themselves. And it's been, I think, a very powerful symbiotic relationship for us to actually be adopting these tools and, like, transforming our own organization, which was started way, way pre-AI into a super AI-native organization ourselves, and Pete has really been leading the charge for that. And so I'm really excited about this episode because I've actually been wanting to talk publicly about all the stuff that we've built internally, and this is the first time that we're doing it. So Pete, perhaps to start off, can you sort of go back to the beginning and, like, talk about, like, there was a pr- a particular, like, moment when we really started adopting these AI tools internally. It was really you who got us

3. 2:15 – 5:07

   The Finance Team Problem That Started It All

   1. JFJared Friedman2:15

      started down that path.

   2. PKPete Koomen2:15

      Sure. Uh, happy to, happy to tell the story here. And it's... I, I like framing it that way because it was a project that I and, and a few engineers got started about a year ago, maybe a little more, but that has since snowballed into just a whole infrastructure layer that's made it possible for us to use AI internally at YC in lots of different ways. And, and that's actually been one of the neatest parts about this, is watching the whole engineering team and, and many partners also just dive in and contribute to this, this infrastructure layer. We started building our own harness inside of YC for kind of YC-specific agents a- about a year ago. And, uh, the original impetus for the project was some of the work that I and a, and a few of the software engineers at YC were doing with our finance team. Just for a bit, a bit of backstory, so YC has, for as long as it's existed, as far as I'm aware, run mostly on our own software, in this era just given us a huge advantage, right? And so with that context, back to this, this moment maybe a year ago, we were sitting down with the finance team talking through a set of tools that we were going to build for them, uh, just to help them run through some of their finance workflows, booking journal entries, uh, logging priced rounds, like all the sorts of things that, that make YC run really. I was seeing kind of two things at once. Like o- on one hand, uh, we, you know, we had this sorta loop going internally, right, where we'd sit down with the finance team. The finance team would describe to our software engineers how, you know, this complicated financial workflow worked, and then software engineers would go and build some purpose-built software where there was a deterministic workflow encapsulating everything that they had been told, and then hand it back to the finance team and so on. And it felt really inefficient. And then at the same time, this was right around the time when agentic tools were really... agentic coding tools were really catching hold, right? And so you had, uh, kind of the first generation, uh, Windsurf and Cursor that were well-established by this point. I think this was right around when Claude Code was, was introduced. And it felt like this was giving me superpowers, right? Um, and then kind of watching this sort of old classical way of building software in YC and then watching how I was doing things on my own machine, this... it just felt like a bigger and bigger divide between those things. And so the original impetus was, "Why don't we try to build some tools at YC that we could use to run agents that would give the finance team control over their own software?" Right? Like, remove the software engineers from this crazy loop where they have to sort of understand these complicated workflows, and give the finance team the tools that they could use to encode their own workflows, not, not as, you know, not as Ruby, uh, but as, as English with prompts, right?

   3. GTGarry Tan5:04

      I mean, what's interesting is, like, uh, we all funded companies

4. 5:07 – 7:20

   SQL Access Changes Everything

   1. GTGarry Tan5:07

      may- maybe even, like, two or three years ago when LLMs were out, but, like, agentic coding wasn't a thing yet. And so the first thing actually was not agentic coding. It was LLMs for writing SQL queries.

   2. PKPete Koomen5:18

      Yes.

   3. GTGarry Tan5:19

      So that's what I remember from, like, the first versions of what you built, was, uh, how, like, good it was and how basically it rhymed with, like, these other failed startups that we had funded. Like, each of us probably funded one-

   4. PKPete Koomen5:32

      [chuckles]

   5. GTGarry Tan5:32

      ... at some point. You know, here it was. It was working, and it worked so well that non-technical people, uh, granted very smart people from finance but with no engineering background, could use these tools to ask real questions.

   6. PKPete Koomen5:44

      I was really surprised too, to be honest. And, and so that... We started with this kind of purpose-built thing for finance and then rewrote it to be more of a general agent loop.Right? And it, and it's, this is now, you see these all over the place now. But, um, I- the first kind of magical moment that I had was we had this l- agent loop, and we had a tool registry, a shared tool registry for kind of YC-specific tools. And the first tool that really was an unlock for me was, I think, a tool, looking back, that you actually built, Jared. It gave these agents the ability to run read-only SQL queries-

   7. JFJared Friedman6:20

      Oh, yes

   8. PKPete Koomen6:20

      ... against our database.

   9. JFJared Friedman6:21

      Yes.

   10. PKPete Koomen6:22

       Right? Um, it was two tools, actually. One was, was running queries against our database, and the other one was the ability to read our model files.

   11. JFJared Friedman6:29

       I remember I built those tools, and I felt a little bit like I was breaking the rules. 'Cause initially, we started with very limited tools that had very z- n- narrowly scoped domains. And I kept getting frustrated because they weren't powerful enough to do the things that I wanted. And so I was like, "What if we just gave the thing, like, access, complete access to the production database where it could just, like, trample on anything?" [laughs] And I sort of, like, surreptitiously pushed it out maybe late at night.

   12. PKPete Koomen6:59

       And it worked.

   13. JFJared Friedman7:00

       [laughs] And it worked, yeah.

   14. PKPete Koomen7:01

       It worked extremely well.

   15. JFJared Friedman7:02

       Yeah.

   16. PKPete Koomen7:02

       Right?

   17. JFJared Friedman7:02

       Yeah. Perhaps foreshadowing, you know, subsequent things like OpenClaw, where it turns out that, like, the thing that was hampering the world was being worried about security and privacy and all the things that could go wrong. And when you, like, worry a bit less, you're like, "Oh my God, these things are unbelievably powerful."

   18. PKPete Koomen7:17

       It's, it's another really good example of this weird split between

5. 7:20 – 9:14

   One Database to Rule Them All

   1. PKPete Koomen7:21

      I'm at work and I'm kind of operating in this really-

   2. JFJared Friedman7:24

      Yeah

   3. PKPete Koomen7:24

      ... narrow box, and I'm at home using Claude Code or, or, or whatever, like OpenClaw or Hermes, and I can do anything.

   4. JFJared Friedman7:31

      Yeah.

   5. PKPete Koomen7:31

      Right? Um, and, and, and trying to, trying to narrow that gap. So why was this so useful, this ability to run SQL queries against our database? Sounds really simple. Well, I think this is where it's important to talk about one of the big advantages that I think YC had coming into this experiment, uh, which is that we run on our own software, and all of that software sits on one Postgres database that has everything that's important to YC's world in it. You know, every company that we funded, there's a companies table. There's a, there's a founders table, right? There's tables for our financial transactions. There's tables for the notes that I leave in our little internal CRM, right? All of these functions that I think a lot of other companies farm out to third-party SaaS tools, we've built our own. And as a result, we have this database with every important piece of context that I can now ask questions like, "Hey, show me all of the investors who invested in a space-related company in the last four batches," right? It just turns out when, when all of that context is in one place, with a little bit of additional, uh, information about how the schema is laid out, an agent can go and ask any, or answer arbitrary questions about, about our business.

   6. GTGarry Tan8:41

      That was a magic moment for sure when I first saw that.

   7. PKPete Koomen8:43

      Yeah.

   8. JFJared Friedman8:44

      And the cool thing for me is that it didn't just make it easier to answer questions. It dramatically increased the number of questions that we would ask, and dramatically increased the, the scale and complexity of the questions that we would dare to ask, where, like, you know, in the, in the old days, back when we were using, like, BI tools, to ask, to ask a question like that, you know, like what investors have invested, like, in space-related companies, that would be, like, several hours of writing SQL. And so, like, unless it was really important, you just wouldn't bother.

   9. PKPete Koomen9:10

      It's just another example of the, you know, the, this

6. 9:14 – 10:07

   Jevons Paradox

   1. PKPete Koomen9:14

      instance of Jevons Paradox that you get-

   2. JFJared Friedman9:16

      Yeah

   3. PKPete Koomen9:16

      ... when you remove the amount of back and forth, uh, between different teams in order to get a thing done, right? If, if in order to ans- ask some kind of complex question about YC, I have to go and knock on s- on, you know, the data science team's door and wait for them to get it through, you know, their backlog, uh, I'm just gonna ask far fewer questions.

   4. GTGarry Tan9:36

      I mean, there are people out there watching this who work in places that still use it. The majority of people live in that world still, and it's 2026, which is a little unfathomable actually.

   5. PKPete Koomen9:47

      There's a long way to go-

   6. GTGarry Tan9:48

      Mm-hmm

   7. PKPete Koomen9:48

      ... I think, which is, which is really exciting.

   8. DHDiana Hu9:50

      I guess one question is how do companies that live in that old world could get sort of wings to move so quickly? Because our, the magic for us was, as you said, everything was, the context was in one place. That made it easy.

   9. GTGarry Tan10:03

      You know, if you think about, um, data science, uh, historically,

7. 10:07 – 12:15

   Denormalizing for Agents (GBrain)

   1. GTGarry Tan10:07

      one of the first things that the Googlers had to figure out was, uh, Bigtable, right? And Bigtable was, you know, instead of schema, you, and joins, you have one big table that, um, can be map reduced.

   2. PKPete Koomen10:20

      Yes.

   3. GTGarry Tan10:21

      And so I think that that's happening again, and I would argue that that's happening now with, um, Karpathy-style knowledge LLM wikis, uh, with GBrain. [laughs]

   4. PKPete Koomen10:31

      [laughs]

   5. GTGarry Tan10:32

      I mean, that's what I'm seeing anyway. Like, you know, obviously I have, I have an OpenClaw. It has-

   6. DHDiana Hu10:36

      Mm-hmm

   7. GTGarry Tan10:36

      ... uh, access to lots of, lots of systems, and then I'm normalizing it to my own schema that's relevant to me and the things that I care about, and it is like denormalization. It's you're taking data and you're putting it into a format that, uh, is more or less optimized for OpenClaw or Hermes Agent, like that particular type of harness to be able to ask questions.

   8. DHDiana Hu10:56

      Mm-hmm.

   9. GTGarry Tan10:57

      And it needs retrieval, it needs RAG, it needs Graph RAG, it needs, uh, you know, hybrid RRF, like there's re-ranking in there. Like, you know, all the things that everyone has learned about retrieval, uh, is now inside GBrain. And then when you give the agents a soul, and it, and you give it, uh, the data, and it knows you and what you care about, like suddenly these things have insane wings. Like I just kind of can't believe how it sees around corners, and you might ask a question, and it'll even, you know, sort of interpret what your question was about and, like, give you a thing that, uh, frankly, like, it would take a human who really knows you well-

   10. PKPete Koomen11:34

       Mm-hmm

   11. GTGarry Tan11:35

       ... to answer.Um, all that's possible now. And so, you know, your question is like all the data is everywhere. My answer from like the OpenClaw Hermes experience with GBrain is like, yeah, you basically have to take that you're gonna denormalize it and you're gonna put it in a format that is optimized for agent retrieval and understanding. You could wrap it in an MCP, but for whatever reason, I just like intuitively I'd be worried. Like it's still sort of, you know, these things are really good at working with MCP and CL- like they're a little even better with CLI. It seems like you have to denormalize and do the Bigtable thing, but, you know, specifically for the agent.

   12. PKPete Koomen12:12

       Looking back over the, the last year and a half, uh, it

8. 12:15 – 14:16

   The Single-Player Era of Agents

   1. PKPete Koomen12:15

      feels like we're still kind of in the single-player era of agents, where the harnesses that have gotten really popular, right, uh, Claude Code, Codex, Pi, OpenClaw, Hermes, they're all designed to be used by a single human running on a single machine. And it makes a lot of sense, right? Because in that environment, these, these agents can do just about anything, right? And they, they make you incredibly powerful. It's, it's, they're a lot of fun to use. I think one of the big problems, uh, that I don't think has been solved well yet by anybody is the multiplayer-

   2. GTGarry Tan12:49

      Mm-hmm

   3. PKPete Koomen12:49

      ... harness, right? It's, it's enabling that kind of superpower, but on a team or an organizational-

   4. GTGarry Tan12:55

      Yeah

   5. PKPete Koomen12:55

      ... level, right? And, and, and that's, I think, been the interesting thing to explore with the infrastructure that we've built at YC, is watching which primitives that we've created that have enabled individuals and teams to use agents. You asked the question about if you're working inside of a kind of a legacy organization, which is like anyone who's more than two years old-

   6. GTGarry Tan13:17

      [laughs]

   7. PKPete Koomen13:17

      [laughs] Uh, what are the things that you can focus on, uh, in order to, to help enable everybody at your org to use AI to, to do more? Uh, and it- we talked about kind of this common context layer, right? And so a data warehouse where just as much of your internal important context lives, it just turns out is extremely useful. There are many tools for connecting individual agent harnesses to, uh, you know, other MCP tools, uh, other, other sources of truth. But l- just like a coding agent inside a monorepo just tends to be much more efficient, watching our agents operating on our single database that has everything in one schema tells me that there's a lot of value at least in getting all of the context into one place. Having an internal tool registry, this is I think the other really important thing that we've built. So in the beginning, like we were talking about, it was just the whole system was really simple. It was like an agent loop and a simple tool

9. 14:16 – 16:24

   350 Tools and a Shared Registry

   1. PKPete Koomen14:16

      registry and, you know, a few other pieces, right? Like a model router underneath. The tool registry is where most of the like YC specific stuff lives, right? The tool registry is what turns these agents into something that's useful at work, and we had like 20 tools at the beginning, including this magical ability to query our, our SQL database. But over time, teams have added more and more tools. Every time we kind of come upon some piece of work at YC, uh, that we think could be improved with an agent, we can just add tools, and there's more than 350 today. I just checked, right? Every team is adding their own tools. You know, I can do things like manage my office hours. Our finance team can, uh, you know, can book journal entries, right? We can help manage the events that we run. Uh, there's tools for all of the important work that we do at YC. And now once these all exist in, in, in, in one place, you can make them available to these internal agents that we've built, but you can also make them available to Claude Code, you know, running, running on, on, on, on our individual machines. So those things above all, I think were the important pieces that we built that if I were working in any other organization, I would focus on building.

   2. GTGarry Tan15:26

      I mean, honestly inspired by what you guys would- did with tools, like this idea of Skillify in OpenClaw, and then actually the most import- the last part of Skillify. Skillify is like this meta skill that I made in OpenClaw where it's like you just do anything in, uh, OpenClaw and Hermes. Hermes actually already has Skillify. They call it something else. Like it makes skills automatically. But the most important thing I, I think is actually like plugging it into the resolver, which is like your agents.md with like the list of things that the agents can do, and then like it links to the markdown entry point that like lets you use a tool basically.

   3. PKPete Koomen16:01

      Mm-hmm. Mm-hmm.

   4. GTGarry Tan16:01

      And so like this thing keeps coming up in all these different contexts, like Claude Code has a skill. The skill registry in Claude Code is actually a resolver. Our tool registry is actually a resolver.

   5. PKPete Koomen16:13

      Mm-hmm.

   6. GTGarry Tan16:13

      And then the weird thing that you have to do on top of that is actually, um, I have a meta skill called Check Resolvable that I call all the time. So I'm always like, I do something that's new or different

10. 16:24 – 18:23

    Skillify, DRY, and MECE Resolvers

    1. GTGarry Tan16:24

       in, uh, in my agent.

    2. PKPete Koomen16:26

       Mm-hmm.

    3. GTGarry Tan16:27

       And then after it does it and I like it, I say Skillify it, and then it becomes basically like a tool call or method call. And then I run Check Resolvable, which is like, you know, look at all of the other skills and, uh, tools that exist and is it, you know, DRY, don't re- don't repeat yourself, and is it, uh, MECE, which is, you know, I'm embarrassed to say a McKinsey term for, um, the consultants use it for, uh, making really good slide decks, uh, mutually exclusive, collectively ex- exhaustive.

    4. PKPete Koomen16:56

       Mm.

    5. GTGarry Tan16:56

       That's like how you're supposed to do slides if you're a McKinsey consultant. But it's useful because it's like an additional layer on top of don't repeat yourself DRY.

    6. PKPete Koomen17:05

       Hmm. Hmm.

    7. GTGarry Tan17:05

       And like the models just seem to know what those things are. And so if you have a DRY and MECE resolver table anywhere, it's actually like the optimal resolver. Like it's bad to have 10 skills that do all the same thing.

    8. PKPete Koomen17:20

       Mm-hmm.

    9. GTGarry Tan17:21

       It's good to have one skill or one tool that has parameters that then let you call them.So I don't know. I think it's like this is like the wildest time to be alive as like an applied computer scientist because it's like simultaneous like discovery of the same useful applied concepts over and over again. And I wonder if like when people were, you know, developing the first versions of Unix or something, it's like discovering a stack and a heap.

    10. PKPete Koomen17:45

        [laughs]

    11. GTGarry Tan17:45

        It feels like we're right at that moment-

    12. PKPete Koomen17:47

        Yes

    13. GTGarry Tan17:47

        ... today. Like we're just coming up with the new primitives for what an agentic system actually is, and you can see it in the parallel sort of development of like we're just trying to do a thing, and it might be in Claude Code, or it might be in our own internal harness, or it might be in OpenClau, it might be in Hermes. Like these things just keep coming back over and over again.

    14. SPSpeaker18:05

        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 25th and 26th to discuss the cutting edge of tech and startups. Apply now for your spot.

    15. PKPete Koomen18:20

        Yeah, it's really interesting to look at how

11. 18:23 – 20:26

    The Self-Improving Dream Cycle

    1. PKPete Koomen18:23

       some of the other companies that are building this stuff, uh, have built their infrastructure because you see a lot of these same primitives in, in each of them, right? Like there's the agent loops, there's tool registries, there's skill registries. Looking at, at, at the way that we're using skills now at YC, so if you, you think of s- skill as a simple abstraction layer over tools, we have a handful of sort of shared skills, uh, that, that we all have access to, uh, through this, through this agent system. And it's been interesting to watch. I think you've talked about this, where this progression of like in the beginning, you were kind of writing your own system prompts, and then skills emerged, and so you started writing your own skills, and then you would start, uh, meta-prompting, where you'd, uh, where, you know, you'd have-

    2. GTGarry Tan19:04

       Do it again

    3. PKPete Koomen19:04

       ... the agent write a skill. Exactly.

    4. GTGarry Tan19:06

       Improve the prompt-

    5. PKPete Koomen19:07

       Yes

    6. GTGarry Tan19:07

       ... automatically.

    7. PKPete Koomen19:08

       Yes.

    8. GTGarry Tan19:08

       Yeah.

    9. PKPete Koomen19:08

       Seeing us kind of do the same progression internally, where we have a couple skills, and now we've gotten to the point where we have these sort of autonomous self-improving loops-

    10. GTGarry Tan19:18

        Yeah

    11. PKPete Koomen19:18

        ... right? Uh, you know, and so, um-

    12. GTGarry Tan19:20

        Auto research from Karpathy again.

    13. PKPete Koomen19:22

        Yes.

    14. GTGarry Tan19:22

        You know? Yeah.

    15. PKPete Koomen19:23

        Yeah.

    16. GTGarry Tan19:23

        Or/goal now in Codex. Like they've, they've incorporated it too.

    17. PKPete Koomen19:26

        We have this general agent that every night will go and read through all of the agent conversations that employees have had and look for, uh, things it could have done better-

    18. GTGarry Tan19:36

        Amazing

    19. PKPete Koomen19:36

        ... and pieces of context that if it had up front, it would have done more efficiently.

    20. GTGarry Tan19:41

        This is OpenClau's dream cycle.

    21. PKPete Koomen19:42

        Yeah.

    22. GTGarry Tan19:43

        And GBrain also has a dream cycle. This is a, um, um, a skill improvement dream cycle, but it could also potentially, um, read all the transcripts and then write them back into the internal, uh, DB, into the internal CRM on like what we know about people and companies.

    23. PKPete Koomen20:02

        Indeed. A- and w- we, we... There are cool examples of using transcripts actually to make these skills more effective as well. One of the shared skills that we have, uh, is a skill that, that partners at YC use to help our companies, uh, write what we call two-sentence descriptions, right? Everybody here has written hundreds of these.

    24. GTGarry Tan20:23

        We should probably explain what a two, two-sentence description actually

12. 20:26 – 23:06

    The Two-Sentence Pitch Skill

    1. GTGarry Tan20:26

       is.

    2. PKPete Koomen20:26

       Sure.

    3. GTGarry Tan20:26

       Yeah.

    4. PKPete Koomen20:26

       So a two-sentence description is a concise way of explaining what your company does in natural language that anyone will understand and why it's interesting.

    5. GTGarry Tan20:35

       Sounds easy, but it's surprisingly hard for founders to actually do this.

    6. PKPete Koomen20:37

       Yes.

    7. GTGarry Tan20:37

       Yes.

    8. PKPete Koomen20:37

       And also no one does it, weirdly. Weirdly, like even the most experienced founders like forget because they have perfect context. Act- interestingly, uh, I now realize YC itself is a, a context engineering, uh, sort of process in that like f- people... We're frequently teaching people, "You have perfect context about what's going on in your brain, but great communication is replicating that same context in someone else's brain," and that's what a two-sentence pitch is. Like, what is it? Like, I don't even know what the heck this is. And then second part is like, is it interesting or valuable? What... You know, is it worth my time? And so that, you know... When I, when I teach two-sentence pitches, that's my favorite way to do it, is like, do I even know what the heck this is?

    9. GTGarry Tan21:21

       Yes.

    10. PKPete Koomen21:21

        Because if you don't know what it is, you can't even ask a question about it.

    11. GTGarry Tan21:24

        Yes.

    12. PKPete Koomen21:24

        It's like something about computers, I guess, whatever. What, what time is lunch again? And then the second part is equally important, which is like, if I've heard that, you know, there are like 20 companies, like there are five other companies in this room that do X. Like, and then I don't understand like why this is noteworthy.

    13. GTGarry Tan21:41

        Yes.

    14. PKPete Koomen21:41

        Like again, I'm like thinking about my pastrami sandwich again, right?

    15. GTGarry Tan21:45

        [laughs]

    16. PKPete Koomen21:45

        So, so the two-sentence pitch like viscerally is important for founders.

    17. GTGarry Tan21:50

        And it's, it's a, it's a simple kind of atomic thing that every partner at YC has practiced over and over and over again. I think Tom, uh, one of the, one of the partners here, wrote a skill that teaches an agent how to, uh, take some context about a company and con- and condense that into a two-sentence description. And so that was his h- sort of handwritten prompt or skill, a- about how that was done. And one of the cool things that happened in the last month or two was that a couple of the other partners took a meeting that they had with a, a group office hours they had with a bunch of the companies in the spring batch and just went through and had every founder try, uh, their hand at, at a two-sentence description and kinda gave them feedback and input. And so kind of the knowledge that lives in a partner's head about how to do this effectively was exchanged back and forth, right, and, and, and now lived in the context of, of that meeting transcript, and handing that back to the agent and saying, "Given, you know, what you've learned by reading through this context, improve the two-sentence description skill." And they got noticeably better-

    18. PKPete Koomen22:59

        Wow

    19. GTGarry Tan22:59

        ... after that. Like this thing is now better than I am, I would, I would argue, at writing those.

    20. PKPete Koomen23:02

        This is how super intelligence happens inside organizations.

    21. GTGarry Tan23:05

        [laughs]

13. 23:06 – 25:10

    How Super Intelligence Compounds

    1. PKPete Koomen23:06

       I mean, this two-sentence pitch thing sounds like something kind of small, but, uh, embedded in it is actually something very powerful. I'm sure you guys have heard, um, Jack Dorsey talk about what he's doing with Block. He basically is trying to turn Block into a mini AGI around helping people in the world make payments to one another, right? Uh, and then this is actually-

    2. GTGarry Tan23:28

       the micro mechanism by which he's going to do that, right? Like, you can look at the operation of any organization as, uh, the aggregate of, you know... I mean, the two-sentence pitch at YC is, that's sort of one of, like, thousands of things that I would argue we do for founders. But, you know, we just walked through a very concrete way where someone wrote a prompt, used it, used a bunch more, other people used it, uh, a bunch of artifacts came off of that around literally, like, the transcript of using it becomes a thing that can be used to meta-prompt and improve, in an automated fashion on a daily basis, the operation of that one skill.

    3. PKPete Koomen24:09

       Yes.

    4. GTGarry Tan24:09

       And then suddenly, that one skill, you just said it, that skill is now better than any of us individually than bef- you know, when before we actually had access to that. And so this is, like, a particular, like, n- needle pin prick in the fabric of, like, how any organization does things. And then how do you build super intelligence inside a company? You do that on everything you do, and it's not more complicated than that. Like, you literally just compose everything that you do, and any given thing that any given person can do, you combine that in aggregate and in this particular process, and, like, you have a super organization.

    5. PKPete Koomen24:47

       Right.

    6. GTGarry Tan24:47

       It's possible now. Like, every single person watching this can do this at any company, at their own company. They can do it at their job. I mean, the interesting thing is that's why you should start a startup, 'cause people are gonna be trapped in organizations with people running organizations that are very powerful and have all these resources and all this capital that do not believe what we just said.

    7. DHDiana Hu25:07

       Because they keep all the context locked down.

    8. GTGarry Tan25:09

       Right.

14. 25:10 – 27:10

    Recording Everything as a Building Layer

    1. GTGarry Tan25:10

       Because it's unsafe.

    2. DHDiana Hu25:11

       It's unsafe. This is one of those things that we talk about, um, how do you build that AI-native organization, right? Part of the key thing is not to ju- just use AI as a copilot. I think that's very 2023, '4, right? This is the, the thing where you use it as a really, the, the building layer for everything, and you need to start recording all the artifacts. Like, people wouldn't have thought of, uh, meeting recordings, and it is one of those reasons why all these, uh, meeting recorders have been taking off. People have been finding them with coaching them on the meetings, but it's not just that. You could take that and improve all the output for you that you do, like writing emails, communication, planning. You have the whole context of everything.

    3. PKPete Koomen25:56

       It's funny you say that. I remember the Dario essay where it's like there's some of the blockers and just the rate of progression of AI are not technical. They're just sort of like social, cultural things. I think it's kind of like a really interesting example. Two years ago, it would've seemed... I just remember it, like, felt odd to just, like, record a meeting or-

    4. DHDiana Hu26:11

       Yeah

    5. PKPete Koomen26:11

       ... like, there was just, like, people trying to figure out what the, like, social etiquette around it was and, like, how intrusive it was. And today, I just feel like it's almost, like, default assumed that, like-

    6. GTGarry Tan26:21

       Yeah

    7. PKPete Koomen26:21

       ... most meetings are being rec- especially if they're on Zoom, but just in general, like, everyone's started recording things now.

    8. PKPete Koomen26:27

       It's a little scary, but I think if you frame this as a way for everyone in an organization to get better at what they do using the, like, collective skill-

    9. GTGarry Tan26:37

       Mm-hmm

    10. PKPete Koomen26:37

        ... and instinct of, of the people they work with, it's incredibly powerful. Having a canonical two-sentence description skill is not just a way to, like, generate a s- a snippet of text for a founder. It's a way to help me get better at understanding what makes for effective founder communication, right? Because now I can tap into everything that Diana and Harj and you two have learned over the many years you've done this job, which are now kind of baked into this skill through the conversations that you've had.

    11. GTGarry Tan27:06

        It's like a shared organizational brain.

    12. PKPete Koomen27:08

        Yes.

    13. DHDiana Hu27:09

        It is very empowering.

    14. GTGarry Tan27:09

        It's like, it's like the

15. 27:10 – 29:18

    The Shared Organizational Brain

    1. GTGarry Tan27:10

       closest thing to us being able to, like, connect our brains.

    2. PKPete Koomen27:12

       Yes.

    3. GTGarry Tan27:12

       Right? Yeah.

    4. PKPete Koomen27:12

       It, it, it totally is, right? And I, I can have an agent now come, and I can do practice sessions with it, right? And I can have it critique my... Like, there, there are so many possibilities once you get all of this knowledge into a place where an agent can, can work with it. Uh, it's a, it's a, it's a very empowering thing for every human in the organization.

    5. GTGarry Tan27:32

       There's some subtle interesting things around here that, like, you know, other people might get wrong that, like, I feel like we've gotten right. I mean, one of them-

    6. PKPete Koomen27:39

       Mm-hmm

    7. GTGarry Tan27:39

       ... is by default, the agent conversation is actually g- um, globally vi- viewable-

    8. PKPete Koomen27:44

       Yes

    9. GTGarry Tan27:44

       ... by any full-time employee at YC. You know, we sort of weren't sure about-

    10. PKPete Koomen27:49

        Mm-hmm

    11. GTGarry Tan27:49

        ... that decision. I mean, it felt right, and it felt like living in the future, but it did not come easily.

    12. PKPete Koomen27:56

        Yes.

    13. GTGarry Tan27:56

        I feel like we had a lot of conversations about, like, "Well, then everyone sees everything. Is that okay?" And like, you know, "What is not okay?" And then I'm glad we made the choice to keep it open, actually, 'cause-

    14. PKPete Koomen28:07

        I agree

    15. GTGarry Tan28:07

        ... people learned how to use it from watching how other people used it.

    16. PKPete Koomen28:10

        Yes. We used that transparency to solve several problems at the same time. One, every agent conversation, as you mentioned, was broadcast internally to a Slack channel, and anybody could join that Slack channel and look and learn, right? And I remember, uh, this is another kind of big unlock moment, was when you started using it really heavily. You were, like, super creative with, with the things you were doing with it, and a lot of us watched that and was like, "Oh, wow, I didn't even occur to me-

    17. GTGarry Tan28:36

        You can do that now. Yeah

    18. PKPete Koomen28:37

        ... to, to, to use it that way," right? It allows you to be a little more lenient on internal security, right? One of the things we talked about earlier was this trade-off where these agents are at their most powerful when they are given unrestricted access to lots of context, which runs counter to the way mor- most organizations work. It turns out that by defaulting to public broadcast for these conversations, you kind of institute a bit of a social control on what people can do with it, uh, that as we learned, I think has been, like, reasonably effective, uh, inside of this high-trust environment at keeping private information private.

    19. GTGarry Tan29:15

        Yeah, what's interesting is, um, it, it betrays two

16. 29:18 – 30:44

    Trust-Default Culture as a Requirement

    1. GTGarry Tan29:18

       traits of, uh, truly agentic, like, 1000x super intelligent organizations-

    2. PKPete Koomen29:23

       Yeah

    3. GTGarry Tan29:23

       ... that I would not have necessarily guessed would exist but are now, like, must exist. If you want to create this type of organization, you have to be relatively egalitarian, and you also have to be trust by default.And then neither of those things, uh, actually are most organizations in the world. If you're the founder of an organization, you actually have to have those at the core of what you're doing.

    4. PKPete Koomen29:44

       And I think, like, that kind of environment honestly works best at startups, right? When it's a small group of people that are a- all aligned and, and, and, and operating in a high-trust environment.

    5. GTGarry Tan29:54

       The other thing you have to do is be willing to spend, like, 10 to $100,000 a year-

    6. PKPete Koomen29:58

       [laughs]

    7. GTGarry Tan29:58

       ... on tokens. But if you're willing to do it, and you invest in the skills, and you, like, actually do everything in an open way with your team that way, like, basically what I realized is it allows you to live in 2028, right? Like, what you spend 100,000 or a million dollars a year on now, it will be commonplace, like, in, in two years, right? It'll... It won't cost 100,000 in a year. It'll cost 10,000, and the year after that, it'll be, like, a couple hundred bucks, right? And everyone will do it. And we'll call it, like, "This is how companies are now." So basically, there's a one-time time warp where you can leapfrog every incumbent, all Fortune 500s, all startups that exist by doing this.

    8. JFJared Friedman30:41

       Like, I'm imagining in the '90s, I wonder if it felt similarly

17. 30:44 – 32:35

    Raising the Floor for New Employees

    1. JFJared Friedman30:44

       when companies started buying computers for their employees.

    2. GTGarry Tan30:47

       Yeah.

    3. JFJared Friedman30:47

       They were probably very expensive, and probably only certain companies really invested in buying these, like, expensive, flaky computer systems for their employees. But, like, what a superpower-

    4. GTGarry Tan30:57

       Yeah

    5. JFJared Friedman30:57

       ... to have a computer when your competitors, like, don't have computers.

    6. DHDiana Hu31:00

       I think more tactically how I've seen this affect, uh, YC has been raising the-

    7. JFJared Friedman31:05

       Raising the bar

    8. DHDiana Hu31:05

       ... the floor.

    9. JFJared Friedman31:06

       Bar.

    10. DHDiana Hu31:06

        The floor, in a sense. What I mean by that is that you could have a new employee joining, and maybe it would've taken them six months to ramp up, but with this, it's sort of like they automatically get a lot of the context from the company working, and they know how the best people and the star players in the or- organization do things by apprenticeship automatically with AI instead of, uh, because partner time is expensive or sometimes the best people in a org, they're very busy, right? And you get to kinda run the simulation of what it's like to be like Pete when he does, like, an awesome job coaching founders on sales or like Garry when he's, like, talking to founders and giving very specific advice. I think it helps all the new, new entrants in the or- organization just be a mini version of you a lot faster.

    11. PKPete Koomen31:52

        One of the first things that I appreciated about being able to use a coding agent was that all of the dumb questions I was too embarrassed to ask, I had no trouble asking, asking the agent. And it... This is kind of that same thing, but at an organizational level, right? You're a brand-new employee. You're embarrassed to ask. You don't wanna bug Harj with a, with a question. And now you don't have to, right? You... And, and which on net means a lot more questions get asked and answered, and people ramp up much more quickly.

    12. JFJared Friedman32:20

        After you had built all of this agent infrastructure at YC, it inspired you to write this essay, Horseless Carriages, that went, like, pretty viral on the internet. Maybe you can, like, explain the ideas behind Horseless Carriages. I think they're still very relevant now.

    13. PKPete Koomen32:32

        It was a critique of a lot of the, the

18. 32:35 – 34:24

    Horseless Carriages

    1. PKPete Koomen32:35

       AI software that I saw being built at the time. And to be totally honest, I think a lot of it still falls into this category.

    2. GTGarry Tan32:41

       It's still like that.

    3. PKPete Koomen32:42

       Yes.

    4. GTGarry Tan32:42

       Yeah, it didn't change.

    5. PKPete Koomen32:43

       Yes. I just saw a lot of examples of, uh, companies building software and adding AI features by sort of slotting a little bit of AI inside of a lot of software, right? And, and the example that I used at the time was the, the kind of email writer that, uh, the, the Gmail, uh, team had, had shipped. But the, the real idea underneath was just kind of that, the, the, the potential for AI is to shift control of software from the developer to the user, right? And, and the, the simple example I started with was basically that all of these kind of like AI as a little feature kept a bunch of prompt context about how the AI should do a job locked away and hidden from the user, which was just this classic example of like, well, it's the developer's job to figure out how all of this stuff should work. So the developer should write that, and we should protect the user from that kind of complexity.

    6. GTGarry Tan33:37

       Safetyism, I hate it.

    7. PKPete Koomen33:38

       Right. And, and, you know, and it, and it's just, again, going back to this contrast between watching the way that some of these tools work and what it was like to use a coding agent on my computer that could do anything, right? And feeling, feeling like I, I had superpowers. I think the conclusion that this essay points to is that as we get better at building AI-native software, it's going to look a lot more like the agent wrapping software-deterministic tools rather than deterministic software wrapping an AI, right? And we've done our best to expose that to internal employees with some of these primitives that we've built. Um, but we have a lot, we have a long way to go.

    8. PKPete Koomen34:21

       The chat as the interface, I just feel something... There's, like,

19. 34:24 – 38:50

    Why Chat Is the Best Interface for Agents

    1. PKPete Koomen34:24

       uh, things going around right now about how there's a need to build a new interface for, like, AI, and what does that look like? And I think that just comes from people who haven't, like, touched and felt it yet.

    2. DHDiana Hu34:33

       [laughs]

    3. PKPete Koomen34:33

       Chat is actually pretty good because, like, you trust the agent, and you increasingly trust the agent to do more of the work, and you trust its decisions, and you don't actually need to, like, have too much of a UI to go in and, like, review the things it's, it's doing. I find it-

    4. GTGarry Tan34:46

       It's time for just-in-time software.

    5. PKPete Koomen34:47

       Yeah, basically, right? Like, yes, occasionally you want it to present you, like, maybe, like, a specific view of something, but for-

    6. GTGarry Tan34:54

       And it could make the software and build it as a single-page JavaScript just purposely-

    7. PKPete Koomen34:59

       Yeah

    8. GTGarry Tan34:59

       ... built for you-

    9. PKPete Koomen35:00

       Yeah

    10. GTGarry Tan35:00

        ... at that moment.

    11. PKPete Koomen35:01

        Yeah.

    12. GTGarry Tan35:01

        And it could be a skill file that could be, like, called any time you want.

    13. DHDiana Hu35:04

        I was thinking a lot about this because I used to be in the camp that, oh, perhap- when ChatGPT came out and it was 2023, that perhaps chat was not gonna be the UI for all these AI applications.

    14. GTGarry Tan35:14

        Mm.

    15. DHDiana Hu35:15

        And I've definitely changed my mind. Part of it is, like, after experiencing all these tools, and I think the more I reflect upon it, why chat is probably the better interface is because it's the closest thing to human language, and human language and writing is basically the closest thing to expression of thinking.

    16. GTGarry Tan35:32

        Mm-hmm.

    17. DHDiana Hu35:32

        So chat is the closest stepping stone to clear intelligence.

    18. GTGarry Tan35:36

        Yeah.

    19. DHDiana Hu35:37

        So you can't justput it in a box. I think it just constrain us too much to have a very specific box. So that's why I thought, I was like, "Okay, all in with chat interfaces." I used to be in the other camp, and it's like-

    20. PKPete Koomen35:49

        And it's multimodal. I know we've talked about, like-

    21. DHDiana Hu35:51

        Yeah

    22. PKPete Koomen35:51

        ... Telegram is not ideal, but I actually really-

    23. GTGarry Tan35:53

        It's pretty good.

    24. PKPete Koomen35:54

        Yeah, it's pretty good. Like-

    25. GTGarry Tan35:55

        I mean, the voice memos, sometimes when I don't want to type, you just do the voice memo, and it's, it feels like I'm talking to you, you know?

    26. PKPete Koomen36:01

        Like, I can give my OpenClaw, like, I can give it text-

    27. GTGarry Tan36:02

        Yeah

    28. PKPete Koomen36:02

        ... I can give it voice, I can give it pictures of things, like-

    29. GTGarry Tan36:05

        It's great

    30. PKPete Koomen36:05

        ... I can give it files. Like, it's, like, pretty good.
20. 38:50 – 40:49

    Just-in-Time Software

    1. PKPete Koomen38:50

       YC or, or tools that others have built, tend to be very small and just add kind of the smallest amount of code ahead of time that you need in order to let the model shine.

    2. GTGarry Tan39:04

       Mm-hmm.

    3. PKPete Koomen39:05

       And you can build an awful lot with that, right? I can write tens of thousands of lines of code, uh, like, like you're saying. But the ability to start at this, like, extremely simple thing that I need to, to understand very little in order to use is incredibly powerful, and I think that's... I think most software in the future is going to look like that.

    4. GTGarry Tan39:23

       We were talking about this earlier, but I think that is what OpenCall did really well. Like, there were, like, a few things that you want. You wanted, like, some ability to give it a bit of personality. You wanted it to, like, persist and last for a long time and have some concept of memory. And it's not, like, perfect, but that's, like, actually, like, good enough as, like, for that use case.

    5. PKPete Koomen39:41

       Mm-hmm.

    6. GTGarry Tan39:42

       Now, Claude Code 2, every time Boris comes and speaks at YC, he spoke with Diana ear- earlier this week. One of, one of the things that really stands out is how obsessed he is with simplicity and with just, like, making the product as small as possible.

    7. PKPete Koomen39:55

       My favorite example of this is, is, uh, the, this open source harness called Py, which is a-

    8. GTGarry Tan40:00

       Yep. That's what, that's what OpenCall uses-

    9. PKPete Koomen40:02

       Yes

    10. GTGarry Tan40:02

        ... as its out of the box, uh-

    11. PKPete Koomen40:03

        Exactly

    12. GTGarry Tan40:03

        ... coding agent.

    13. PKPete Koomen40:04

        It's this beautiful piece of software which is just, like, the smallest possible coding agent. You can use Py to ex- modify and extend Py, right? And it's this kind of idea of, like, self-extending and self-referential software. It's really fascinating. Uh, and you're right, OpenClaw was built on top of that. One of the things I'm very curious to see is how many other sort of pieces of classic software emerge in this form as this kind of minimal thing that you start with, uh, and then use an agent to extend over time. I think more and more s- I mean, looking at, honestly, the benefits that we've gotten from having our own customizable software, I suspect that a lot of-

    14. GTGarry Tan40:40

        Yeah

    15. PKPete Koomen40:40

        ... commercial software, uh, will come with this capability-

    16. GTGarry Tan40:44

        Yeah

    17. PKPete Koomen40:44

        ... uh, out of the box in the future.

    18. GTGarry Tan40:45

        There's a really interesting subtle thing that I wanted to talk about around,

21. 40:49 – 43:32

    Centralizing vs. Decentralizing AI

    1. GTGarry Tan40:49

       like, what I learned from your essay, uh, which is, like, AI can either be centralizing or decentralizing. And, um, the Google Gmail, like, I can't change the prompt thing is, like, the perfect example of that. We basically have a choice to be made over the next... I don't think it's even that long. I think it's, like, 18 to 24 months. It might take five years. But, um, there are sort of two scenarios, and, uh, what comes to mind is literally, like, the, uh, 1984 Macintosh commercial by Apple-

    2. PKPete Koomen41:18

       [laughs]

    3. GTGarry Tan41:18

       ... where it's like, is 2034 going to be like 1984? And, you know, the 1984 case would be we have centralized control, like, there are five kings. There's only, you know, one of them maybe wins. They have the most advanced AI. They have, uh, end-run around all compute and power. They have all the space data centers 'cause they c- you can't build any terrestrial data centers in America anyway. There's this, like, centralization of control, and not only that, they don't let you run your own prompts.Like, they literally do the Gmail thing, but, like, for your whole computing existence, right? And this would be as if, like, personal computers never existed and there were only mainframes and mini computers. Like, this is sort of lost to the sands of time, but, you know, in the 1960s and '70s when computers first came out, like, you couldn't go to the store like you can today. You couldn't go to an Apple store and just buy an iPhone, let alone, uh, a Mac. You had to get access to, like, this thing that was worth, like, hundreds of thousands of dollars to millions of dollars, and it was only-

    4. JFJared Friedman42:22

       And it was like, and it was, like, tightly locked down-

    5. GTGarry Tan42:24

       Yeah

    6. JFJared Friedman42:24

       ... by corporate policies. You're right.

    7. GTGarry Tan42:26

       Yeah.

    8. JFJared Friedman42:26

       And the, and the thing that really spurred the computing revolution was when people started having personal computers that, that they could experiment on.

    9. GTGarry Tan42:33

       Yeah, and it was just like the priesthood, right? There was a small priesthood and an institutional base that controlled capital, literally the means of production. And so, you know, this is, like, a coherent future that we could live in that I don't want to live in, and the alternative to that is actually, uh, embedded in the Homebrew Computer Club. It's embedded in the revolution that Steve Jobs and Steve Wozniak gave us when they were in the garage in Mountain View-

    10. JFJared Friedman43:00

        Mm-hmm

    11. GTGarry Tan43:00

        ... literally soldering together breadboards, and they, like, sold 500 of these Apple 1s. And I think we're at the Apple 1 moment right now. We are coming up with the primitives. We are learning how do these things work and how do we sell it and how do we package it? Uh, but then there's, like, a lot of choices right now, right? Like, most people, the mass- you know, a billion users use ChatGPT, and ChatGPT, like, gives you a little access, but MCP is really locked down. You actually, you know, can't hook things up

22. 43:32 – 46:28

    The Personal AI Revolution

    1. GTGarry Tan43:32

       to your own databases that easily, um, and, you know, for what? Safety? Like, I would argue Claude is, like, a little bit more open, but not really. Perplexity Computer is probably the best version of it, but it's still, like, you know, d- pretty limited compared to what you could do with OpenClaw and Hermes Agent. And so what does the, uh, revolution look like that is, like, the true personal AI moment? And that's what I hope that we are building with things like GBrain and, you know, H- Hermes Agent and OpenClaw, like the ability to run your own software, to change your own prompts, to test all of it, to have your own private repo that, like, you know is only yours, um, to be able to choose which model to use, and maybe it's an open weight model. Like, to me, that's sort of the white pill for AI is, uh, we could have corporate control, no control of your own prompts, and, like, literally the AI happens to you. [laughs] You know, you're under the API line. Or, like, there's this other alternative where I want, like, a billion people to actually control and program for themselves what are these things. This should be an extension of yourself and what you care about, not what, you know, Meta or Alphabet or even OpenAI or Anthropic care about.

    2. PKPete Koomen44:51

       I always really bristle when I see AI framed as a way to replace people because it just doesn't match the way that I have experienced it and the way that so many of the people around me have experienced it, not as a replacement for humans, but as a thing that empowers. If you look at, at, at kind of how tech has developed since the era of, of mainframes to PCs to the internet, which gave everyone, like, a publishing platform to w- like, it's, it's a story overall, above all, of individual empowerment, and I think AI, uh, is going to play out the same way. I think it is going to enable us to do more, uh, than we could before. I think it's going to eliminate kind of the drudgery style work that, like, made a lot of my job painful in the past.

    3. GTGarry Tan45:33

       To me, it's like we have to make choices to do so.

    4. JFJared Friedman45:36

       Mm-hmm.

    5. GTGarry Tan45:36

       By default, like, a company is not open. By default, a company is, uh, command and control. By default, maybe the leadership gets access to these tools, but, like, the, you know, line level people, the staff people don't, right? And, like, y- we need, like, a radically different type of organization, and we need to actually offer computing in a different way, and, um, these are all choices, and the people who are watching are gonna be the people who build all these things in society. So we better choose well. Well, that's all the time we have for today. I mean, I think we covered some pretty heavy stuff, but Pete, thanks for joining us.

    6. PKPete Koomen46:12

       Thanks.

    7. SPSpeaker46:12

       Thank you.

    8. PKPete Koomen46:13

       Thank you.

    9. GTGarry Tan46:14

       Thanks for watching, guys. We'll see you guys on the next one. [upbeat music]

Episode duration: 46:29