The Agent Network — Dharmesh Shah, Agent.ai + CTO of HubSpot

Hey everyone, welcome back to the Laden Space podcast. This is Alesio, partner and CTO at Deible Partners, and I'm joined by my co-host Swix, founder of Small AI. Hello! And today we are super excited to have D. Sha join us. Uh, I guess your relevant title here is founder of Agent AI.

Yeah, that's true for this. Uh, yeah, creator of Agent. And yeah, co-founder of HubSpot. But yeah, co-founder of HubSpot, which I followed for many years—uh, I think 18 years now.

Yeah, uh, going to be 19 soon. And you caught—you know, people can catch up on your HubSpot story elsewhere. I should also thank Sean Purri, who's uh, who I've chatted with back and forth, uh, who's been, I guess, getting me in touch with your people, but also I think like just giving us a lot of context because obviously my first million joined you guys, um, and then and they've been they've been chatting with you guys a lot. So for the business side, we can talk about that, but I kind of wanted to engage your CTO, agent engineer side of things. So how did you get Agent AI? Let's see.

Um, so I've been working—I'll take a like a half step back—a decade or so ago, even actually more than that. So even before HubSpot, the company I was contemplating that I had named for was called Innosoft, and the idea behind Innosoft was a natural language interface to business software. Now realize this is 20 years ago, so that was a hard thing to do. But the actual use case that I had in mind was, um, you know, we had data sitting in business systems like a CRM or something like that, and my kind of what I thought clever at the time was, oh, what if we used email as the kind of interface to get to business software? And the motivation for using email is that it automatically uh works when you're offline. So imagine I can like I'm getting on a plane, or I'm on a plane—there was no internet on planes back then—it's like, oh, I'm going through business cards from an event I went to. I can just type things in into uh into an email, just have them all in the backlog when it reconnects; it sends those emails to a processor that basically kind of parses the effectively the commands, um, and updates the software, sends you the file, whatever it is. And there was a handful of commands. Um, I was a little bit ahead of the times in terms of what was actually possible. And I reattempted this natural language thing uh with a product called Chatspot that I did back uh 20—

Yeah, this is your first like post-ChatGPT project. I saw it come out.

Yeah. Um, and so I've always been kind of um, kind of fascinated by this natural language interface to to software because, you know, as software developers, myself included, we've always said, oh, we build intuitive, easy-to-use applications, and it's not intuitive at all, right? Because what we're doing is taking the mental model that's in our head of what we're trying to accomplish with said piece of software and translating that into a series of touches and swipes and clicks and things like that. U and there's nothing natural or intuitive about it. Um, and so natural language interfaces, for the first time, you know, whatever the thought is you have in your head, expressed in whatever language that you normally use to talk to yourself in your head, you can just sort of emit that and have software do something. And I thought that was a kind of a breakthrough, uh, which it has been, uh, and it's gone—so that's where I first started getting into um the the J of AI side of things because now it actually works, right? So once we got ChatGPT and you can take—even with a few-shot example—convert something into structured—even back in the Chat—the ChatGPT 3.5 days, it did a decent job at a few-shot example, convert something to, you know, to structured text if you knew what kinds of intents you were going to have. Um, and so that happened, um, and that ultimately became a HubSpot project, but then Agent intrigued me uh because I'm like, okay, well that's the next step here. So chat—great, love chat UX—but if we want to do something even more meaningful, it felt like the next kind of um advancement is not this kind of I'm chatting with some software in kind of a synchronous back-and-forth model, is that software is going to do things for me um in kind of a multi-step way uh to try and accomplish some goals. So um, yeah, that's when I first got started like, okay, what would that look like? Um, yeah, so and I've been obsessed ever since.

By the way, so which goes back to your first experience with it, which is like you're offline and you do a task; you don't need to do it right now, you just want to queue it up for somebody to do it for you. Yes. As you think about agents, like let's start at the easy question, which is like how do you define an agent? Maybe you mean the hardest question in the universe, is that what you mean?

Um, you said you have an irritating take.

I do have an irritating take. Um, I think—well, some number of people have been irritated, including within uh within my own team—so I have a very broad definition for agent, um, agents, which is: it's AI-powered software um that accomplishes a goal. Period. That's it. And what irritates people about it is like, well, that's so broad as to be completely non-useful, and I and I understand that—understand the the criticism—but in my mind, um, if you kind of fast-forward months, I guess in uh in AI years, um, the implementation of—we're already starting to see this, and we'll talk about this—different kinds of Agents, right? So I think in addition to having a a usable definition—and I like yours, by the way—we should talk more about that—that you just came out with the classification of Agents, actually, is also useful, which is: is it autonomous or non-autonomous? Is it—does it have a deterministic workflow? Does it have a non-deterministic workflow? Is it working synchronously? Is it working asynchronously? Uh, then you have the different kind of interaction modes: is it a chat agent, kind of like a customer support agent would be; you're having this kind of back and forth? Is it a workflow agent that just does a discrete number of steps? Um, so there's all these different flavors of Agents. So if I were to draw it in a Venn diagram, I would draw a big circle that says this is agents, and then I have a bunch of circles, some overlapping, um, because they're not mutually ex—exclusive. And so I think that's what's interesting, and we're seeing uh development along a bunch of different paths, right? So if you look at the first implementation of like agent frameworks, uh you look at you BabyAGI and um AutoGPT—I think it was not AutoGen, that's the Microsoft one—they were way ahead of their time because they assumed this level of reasoning and execution planning capability that just just did not exist, right? So it was an interesting experim—thought experiment, um, which is what it was. Even you—the guy that I'm an investor in—Yoh's—Yoh's fund that did BabyAGI—it wasn't ready, but like it was a sign of what was to come. And so the question then is like, when when is it ready? Um, and so you know, lots of people talk about like the state-of-the-art when it comes to agents. I'm a pragmatist, so I think of like the state of the practical: it's like, okay, well, what can I actually build uh that has commercial value uh or solves actually some discrete problem uh with some you kind of some baseline of uh kind of repeatability or um, you know, verifiability. So yeah, there was a lot and very very interesting. I'm not irritated by it at all. I think as as you know, I take a sort of anthropological view or linguistics view, and in linguistics you don't want to be prescriptive, you want to be descriptive. So you're a goals guy; that's that's the key word in your thing, and other people have other definitions that might involve like delegated trust or non-deterministic work, LLM in the loop, all that stuff. The other thing I I was thinking about like just the comment on BabyAGI, AutoGPT. Yeah, in that piece that you just read, we went—I I was able to go through our backlog and just kind of track the the winter of agents and then the summer now. Y and it it really—it really—you can—we can tell the whole story like as as an oral history just following that that thread, and it's really just like I think I tried to explain the why now, right? Like I had—there's better models, of course; there's better tool use uh with like they're they're just more reliable, better tools with MCMC and all that stuff, and I'm sure you have opinions on that too.

I do.

Business model shift, which you like a lot. I just—I just heard you talk about Ras with uh MFM guys. Cost is dropping a lot.

Yep.

Uh, inference is getting faster; there's more model diversity, which I think is a subpoint; it it means that like you have different models with different perspectives; you don't get stuck in the basin of attraction of a single model; you can just get out of it by just switching models; multi-agent research and re-fine-tuning. So I just wanted to let you respond to like any of that.

Yeah, a couple things. Um, connecting the dots on kind of the definition side of it, so we'll get the irritation out of the way—completely. I have one more even more irritating leap on the agent definition thing. Um, so here's the way I think about it, by the way, the kind of word agent—I looked it up like the English dictionary definition—agent is when you have someone or something that does something on your behalf, like a travel agent or a real estate agent acts on your behalf, is like proxy, uh, which is a nice kind of general definition. Um, so the other direction I'm sort of headed—this is going to tie back to tool calling and MCMC and things like that—is if you—and I'm not a biologist by any stretch of imagination—but we have these uh single-celled organisms, right? Like the simplest possible form of what one would call life, but it's still life; it just happens to be single-celled, and then you can combine cells, over and then cells become specialized over time, and you have much more sophisticated organisms, you know, kind of further down the spectrum. In my mind, at the most fundamental level, you can almost think of having atomic agents: what is the simplest possible thing that's an agent that can still be called an agent? What is the equivalent of a kind of a single-celled um organism? And the reason I think that's useful is right now we're headed down the road, which I think is very exciting, around tool use, right? Um, that says, okay, the LLMs now can be provided a set of tools uh that it calls to accomplish whatever it needs to accomplish in the kind of furtherance of whatever goal is is trying to get done. And I'm not overly bothered by it, but if you think about it, if you just squint a little bit and say, well, what if everything was an agent, and what if tools were actually just atomic agents, because then it's turtles all the way down, right? Then it's like, oh, well, all that's really happening with tool use is that we have a network of agents that know about each other through something like an MCMC and can kind of decompose a particular problem and say, oh, I'm going to delegate this to this set of agents, and why do we need to draw this distinction between tools, which are functions most of the time, and an actual agent? Um, and so I'm going to write this irritating LinkedIn post, uh, you know, proposing this: it's like, okay, uh, and I'm not suggesting we should call even functions, you know, call them agents, but there is a certain amount of elegance that happens when you say, oh, we can just reduce it down to one primitive uh which is an agent that you can combine in in complicated ways to kind of, you know, raise the level of abstraction and accomplish higher-order goals. Anyway, I'd say that—thank you for coming to my TED Talk on what—how do you define the—

Yeah.

Minimum valuable agent? Do you already have a definition for like where you draw the line between a cell and an atom, or—

Yeah.

So in my mind, it has to at some level use AI in order for it to—otherwise it's just software, right? Just like, you know, we don't need another word for that. And so that's probably where I draw the line. So then the question, you know, the counterargument would be, well, if that's true, then lots of tools themselves are actually not agents because they're just doing a database call or a REST API call or whatever is they're doing, and that does not necessarily qualify them, which is a fair counterargument, and I I accept that—it's that's a good argument. Um, I still like to think about—because we'll talk about multi-agent systems because I think—so we've accepted—which I think is true—lots of people have said it—and you've hopefully uh combined some of those clips of uh really smart people saying this is the year of agents, and and I completely agree—it is the year of Agents, but then shortly after that, it's going to be uh the year of uh multi-agent systems or multi-agent networks. I think that's where it's going to be headed next year.

Yeah, OpenAI's already on that.

My quick philosophical engagement with you on this cell thing is—is I often think about kind of the other spectrum, the other end of the cell spectrum. So single-celled is life; multi-celled is life; and you clump a bunch of cells together in in a more complex organism; they become organs like an eye and a liver, whatever; and then obviously we are—we consider ourselves ourselves one life form; there's not like a lot of lives within me; I'm just one life. And now obviously like I don't think people—don't really like to anthropomorphize agents and in AI, but we are extending our our consciousness and our brain and our functionality out into machines. I just saw Urb, which is—Limitless pendant in my pocket, which I got—I got one of these boys.

Yeah.

I'm testing it all out, you know—got got to be early adopters—but like we want to extend our personal memory into these things so that we can—we be good at the things that we're good at, and you know, machines are good at, and machines are there. So like my definition of life is kind of like going outside of my own body now. I don't know if you've ever had like reflections on on that, like how yourself is like actually being distributed outside of you.

Yeah, I don't—

Fancy myself a philosopher.

Uh, but you went there, so—

Yeah, I did go there. Um, I'm fascinated by kind of graphs and graph theory and networks and have been for a long, long time, uh, and and to me we're sort of all nodes in this kind of larger thing; it just so happens that we're looking at individual kind of life forms as as they exist right now, but um, so the idea is uh when you put a podcast out there, there's these little kind of uh nodes you're putting out there of like, you know, conceptual ideas—once again, you have varying kind of forms of those little nodes that are up there and are connected in in sundry ways—and so I just think of myself as being a node in a massive, massive network, and I'm producing more nodes as I, you know, put content or ideas, um, and um, you know, you spend some portion of your life collecting dots—experiences, people—um, and some portion of your life then connecting dots from the ones that you've collected over time, and I found that really interesting things happen uh and you really can't know in advance how those dots are necessarily going to connect um in the future. And that's—yeah, so that's my philosophical—that's uh—

Yeah.

Do you like graph as an agent abstraction? That's been one of the hot topics with LangChain and Pinecone and all that.

I do. Uh, the thing I'm more interested in in terms of uh use of graphs—and then there's lots of work happening on that now—is uh graph data stores as an alternative in terms of knowledge stores and knowledge graphs uh because, you know, so I've been in software now uh 30-plus years, right? So it's not 10,000 hours, it's like 100,000 hours; I've spent doing this stuff. And so I've—grew up with—so back in the day, you—I started on mainframes—there was a product called IMS from IBM, which is basically an index database—what we'd call like a key-value store today—uh, then we've had relational databases, right? We have tables and columns and foreign key relationships; we all know that. We have document databases like MongoDB, which is sort of a nested structure keyed by a a specific index; uh, we have vector stores, vector embedding databases, and graphs are interesting uh for a couple of reasons. One is—so it's not classically structured in a relational way—was when when you say structured database to most people, they're thinking tables and columns in a relational database and set theory and all that. Graphs still have structure, but it's not the tables and columns structure. Um, and you could wonder—and people have made this case—uh that they are a better representation of knowledge for LLMs and for AI generally than other things. Uh, so that's kind of thing number one conceptually, and that might be true; I think is possibly true. Uh, and the other thing that I really like about that in the context of um, kind of data stores for for RAG is, you know, RAG—you say, oh, I have a million documents; I'm going to build a vector embedding; I'm going to come back with the, you know, top X based on the semantic match, and that's fine; all that's very, very useful, but the reality is something gets lost in the chunking process—in the—okay, well, those tend—like you don't really get the whole picture, so to speak, and maybe not even the right U set of dimensions on the kind of broader picture, and it makes intuitive sense to me uh that if we did capture it properly uh in a graph form that maybe that feeding into a a pipeline um will actually yield better results for some use cases. I I don't know, but uh—

Yeah.

And do you feel like at the core of it there's this difference between imperative and declarative programs? Because if you think about HubSpot, yeah, it's like, you know, people—and graph kind of goes hand in hand, you know, but I think maybe the software of before was more like primary foreign key–based relationships versus now the models can traverse through the graph more easily.

Yes.

Uh, so I I like that representation; there's something um just conceptually elegant about graph, and just from the representation of it, they're much more discoverable; you can kind of see it; there's observability to it, um, versus kind of embeddings, which you can't really do much with as a human, um, you know, once they're in there, you can't pull stuff back out. But yeah, I I like that kind of idea of it. And the other thing that's kind of uh—because I love graphs—I've been long obsessed with PageRank from back in the early days, um, and you know, one of the kind of simplest algorithms in terms of coming up uh, you know, with a—and everyone's been exposed to PageRank—but the idea is that—and so I had this other idea for a project, not a company—and I have hundreds of these—called NodeRank, is to be able to take the idea of PageRank and apply it to an arbitrary graph um that says, okay, I'm going to define what authority looks like and say, okay, well, that's interesting to me because then if you say I'm going to take my knowledge store and maybe this person that contributed some number of chunks to the um to the graph uh data store has more authority on this particular use case or prompt that's being submitted uh than this other one, or maybe this one was more popular, or maybe this one has whatever it is; there there should be a way for us to kind of rank nodes in a graph and sort them in in some some useful way. Um, yeah, so I I think that's generally useful for for anything. I think the the problem like uh—so even though at my conferences graph RAG is super popular and uh people are getting Knowledge Graph religion—and I will say like it's getting—space—getting traction in two areas: conversation memory and then also just RAG in general—like the the the document data source—most ML practitioners would say that Knowledge Graph is kind of like a dirty word, and the graph database people get graph religion; everything's a graph, and then they they go really hard into it, and then they get a they get a graph that is too complex to navigate.

Yes.

Um, and so like the the simple way to put it is like you—at running HubSpot—you know the the power of of graphs the way that Google has pitched them for for many years—but uh I don't suspect that HubSpot itself uses a Knowledge Graph.

No.

Yeah, so uh when is it over-engineering? Basic—it's a great question. Um, I don't know. So the question now like in AI land, right, is the—do we necessarily need to understand—so right now LLMs for for the most part are somewhat black boxes, right? We sort of understand how the you—the algorithm itself works, but we really don't uh know what's going on in there um and and how things come out. So if uh a graph data store is able to produce the outcomes we want—it's like, here's a set of queries I want to be able to submit, and it comes out with useful content—maybe the underlying data store is as opaque as uh vector embeddings or something like that, but maybe it's fine; maybe we don't necessarily need to understand it to get utility out of it. And so maybe if it's messy, that's okay; um, that's—it's just another form of lossy compression; it's just lossy in a way that we just don't completely understand in terms of because it's going to grow organically uh and it's not structured; it's like we're just going to throw a bunch of stuff in there; let the the equivalent of the embedding algorithm—whatever they called in graph land—so the one with the best results wins, I think. So yeah—or this is the practical side of me is like, yeah, it's—if it's useful, we don't necessarily understand it. I have—I mean, I'm happy to push back as long as you want—it's not practical to evaluate like the 10 different options out there because it takes time, it takes people, it takes, you know, resources, right? That's the first thing. Second thing is your evals are typically on small things. Y and some things only work at scale.

Yep.

Like graphs.

Yep.

Yep, that's—that's—

Yeah.

No, that's fair. I think this is one of the challenges in terms of implementation of graph databases is that the most common approach that I've seen developers do—I've done it myself—is that, oh, I've got a Postgres or MySQL, whatever; I can represent a graph with a very simple set of tables with a parent-child thing or whatever, and that sort of gives me the ability—why would I need anything more than that? And the answer is, well, if you don't need anything more than that, you don't need anything more than that, but you—there's a high chance that you're sort of missing out on the actual value that the graph representation gives you, which is the ability to traverse the graph uh efficiently in ways that kind of—going through the U traversal in a relational database form—even though structurally you have the data, practically you're not going to be able to pull it out in in useful ways. So you wouldn't—like—represent a social graph uh in using that kind of relational table model; it just wouldn't scale, wouldn't work. I think we want to move on to MCMC, but I just want to—like—just engineering advice. Obviously, you've—you've—you've run uh—you've—you've had to do a lot of projects and run a lot of teams. Do you have a general rule for over-engineering or, you know, engineering ahead of time, you know, like because people—we know premature engineering is the root of all evil, Y, but also sometimes you just have to—

Yep.

When do it?

Yes, it's a great question; this is a a question as old as time almost, which is what's the you know right and wrong levels of abstraction; that's effectively what uh we're answering when we're trying to do engineering. I tend to be a pragmatist, right? So here's the thing: um, lots of times doing something the right way has only a marginal increased cost; in those cases, just do it the right way. And this is what makes a a great engineer or a good engineer better than uh not so great one; it's like, okay, all things being equal, if it's going to take you, you know, roughly, you know, close to constant time anyway, might as well—well, do it the right way, like—so do things well. Then the question is, okay, well, am I building a framework as the reusable library? To what degree uh what am I anticipating in terms of what's going to need to change in this thing, you know, along what dimension? I think like a business person in some ways, like what's the return on calories, right? So uh and you look at um the expected value of—it's like, okay, here are the five possible things that could happen; try to assign probabilities—like, okay, well, if there's a 50% chance that we're going to go down this particular path that someday—like or one of these five things is is going to happen—and it cost you 10% more to engineer for that, it's basically—it's something that yields uh kind of interest-compounding value um as you get closer to the time of of needing that versus having to take on debt, which is when you under-engineer it; you're taking on debt that you're going to have to pay off when you do get to that eventuality where something happens. One thing as a pragmatist—so I would rather under-engineer something than over-engineer it if I were going to air on the side of something, and here's the reason: is that when you under-engineer it, uh, yes, you take on tech debt, uh, but the interest rate is relatively known, and payoff is very, very possible, right? Which is, oh, I took a shortcut here, as a result of which now this thing that should have taken me a week is now going to take me four weeks—fine—but if that particular thing that you thought might happen never actually—you never have that use case transpire—just doesn't—it's like, well, you just save yourself time, right? And that has value because you were able to do other things instead of uh kind of slightly over-engineering it or way over-engineering it. But there's no perfect answer; it's an art form in terms of—and yeah, we we'll bring kind of this layers of abstraction back on the code generation conversation, which we'll—I think—have later on, but I was going to ask—we we can just jump ahead quickly—like, as you think about Vibe coding and all that, how does the percentage of potential usefulness change? When I feel like with over-engineering a lot of times it's like the investment in syntax, it's—about the investment in like architecting, Y.

Yeah.

How does that change your calculus?

A couple of things, right? One is—um, so you know, going back to that kind of ROI or return on calories kind of calculus—or—if you think through—it's like, okay, well, what is it going to cost me to put this layer of abstraction above the code that I'm writing now, anticipating kind of future needs? If the cost of fixing uh or doing under-engineering right now uh will trend towards zero, that says, okay, well, I don't have to get it right right now because even if I get it wrong, I'll run the thing for 6 hours instead of 60 minutes or whatever.

It doesn't really matter, right, like, because that's going to trend towards zero, to be the ability to refactor code, um, and because we're going to, not that long from now, we're going to have, you know, large code bases be able to exist, uh, you know, as as context, uh, for a code generation or a code refactoring, uh, model. So I think it's going to make it, make the case for under engineering, uh, even stronger, which is why take on that cost, you just pay the interest when you get there; it's not just go on with your life. Vibe coded and, uh, come back when you need to.

Yeah, sometimes I feel like there's no decision-making in some things. Like today I built auto-save for like our internal notes platform, sure, and I literally just VI coded it. I don't know if it's over or under engineered; I just VI coded it, and I feel like at some point we're going to get to the point where the models kind of decide where the right line is. Yes, but this is where the, like, the, in in my mind, the danger is, right. So there's two sides to this: one is the cost of kind of development and coding and things like that, stuff that, you know, we talk about, but then, like in your example, you know, one of the risks that we have is that because adding a feature, uh, like a save or whatever the feature might be, to a product, as that price tends towards zero, are we going to be less disc- discri about what features we add as a result of making more product products more complicated, which has a negative impact on the user, negative impact on the business, um, and so that's the thing I worry about. If it starts to become too easy, are we going to be too promiscuous in our, uh, kind of extension, adding product extensions and things like that? It's like, ah, why not add XYZ? Or whatever, back then it was like, oh, we only have so many engineering hours or story points or however you measure things, uh, that at least kept us in check a little bit.

Yeah, and then over engineering, you're like, yeah, it's kind of like you're putting that on yourself. Yeah, like now it's like the models don't understand that if they add too much complexity it's going to come back to bite them later, so they just do whatever they want to do. Yeah, and I'm curious where in the workflow that's going to be where it's like, hey, this is like the amount of complexity and over engineering you can do before you got to ask me, yeah, if we should actually do it, yep, versus like do something else. I think that, so you know, we've already, it's like we're leaving this, uh, in the code generation world, this kind of compressed, um, cycle time, right? It's like, okay, we went from autocomplete in the GitHub co-pilot to like, oh, finish this particular thing and hit tab, to a, oh, I sort of know your file or whatever, I can write out a full function to you, to now I can like hold a bunch of the contexts in my head, uh, so we can do app generation, which we have now with lovable and bolt and repet agent and other things. So then the question is, okay, well, where does it naturally go from here? So we're going to generate products; makes sense. We might be able to generate platforms; it's, oh, I want a platform for ERP that does this whatever, and that includes the APIs, includes the product and the UI and all the things that make for a platform. There's no nothing that says we would stop like, okay, can you generate entire software companies someday, right, uh, with the platform and the monetization and the go-to-market and the whatever? And you know, that that's interesting to me in terms of, uh, you know, what, when you take it to almost ludicrous levels of, uh, abstract, it's like, okay, turn it to 11.

Yeah, you mentioned Vibe coding, so I have to, this is a blog post I haven't written, but I'm kind of exploring it. Yeah, is the junior engineer dead? I don't think so. I think what will happen is that the junior engineer will be able to, if all they're bringing to the table, uh, is the fact that they are a junior engineer, then, then yes, they're likely dead, but hopefully if they can communicate with carbon-based life forms, they can interact with product; if they're willing to talk to customers, they can take their kind of basic understanding of engineering and how, uh, kind of software works. I think that has value. So I have a 14-year-old right now who's taking Python programming class, and some people ask me it's like, why is he learning coding? And and my answer is, is because it's not about the syntax, it's not about the coding; what he's learning is like the fundamental thing of like how things work, and there's value in that. I think there's going to be timeless value in systems thinking and, uh, abstractions and what that means, uh, and whether functions manifested as math, which he's going to get exposed to regardless, or there are some core primitives to the universe, I think, uh, that the more you understand them, those are what, what I would kind of think of as like really large dots in your life that will have a higher gravitational pull and value to them that you'll then be able to. So I want him to collect those dots, and he's not resisting, so it's like, okay, while he's still listening to me, I'm going to have him do things I think will be useful.

Yeah, you know, part of one of the pitches that I evaluated for AI engineer as a term is that maybe the traditional interview path or career path of software engineer goes away, which because what's the point of LeetCode? Yeah, yeah, and, uh, you know, it actually matters more than, you know, how to work with AI and to implement the things that you want. Y that's one of the like interesting things that's happened with generative AI, um, you know, you go from machine learning and the models and just that underlying form of which was like true engineering, right, like the actual what I call real, I don't think of myself as a real engineer, actually; I'm a developer, um, but now with generative AI, we call it AI, uh, and it's obviously got its roots in in machine learning, but it just feels like fundamentally different to me. Like you have the vibe; it's like, okay, well, this is just a whole different approach to software development, to so many different things, and so I'm wondering now like an AI engineer is like, if you like to draw the Venn diagram, it's interesting because the cross between like AI things, J of AI, and what the tools are capable, what the models do, and this whole new kind of body of knowledge that we're still building out, it's still very young, intersected with, uh, kind of classic engineering, uh, software engineering. Yeah, I, I, I just described the overlap; is it separates out eventually until it's its own thing? Yeah, but it's, uh, starting out as a software. Yeah, that makes sense.

So to close the Vibe coding loop, uh, the other big hype now is MCPs. Obviously, I would say Claw desktop and Cursor are like the two main drivers of MCP usage. I would say my favorite, the Sentry MCP, I can pull in errors, and then you can just put the context in in Cursor. How do you think about that abstraction layer? Does it feel almost too magical in a way? Do you think it's like you get enough because you don't really see how the server itself is then kind of like repackaging the information for you? I, I think, uh, MCP as a standard, um, is one of the better things that's happened in the world of AI because a standard needed to exist; absent a standard, there was a set of things that just weren't possible. Now we can argue whether it's the best possible manifestation of a standard or not; does it do too much, do it too little? I get that, but it's like just like simple enough to both be useful and understandable and adoptable by mere mortals, right? It's not overly complicated; uh, you know, a reasonable engineer can put a stand up an MCP, um, uh, server relatively easily. The thing that has me excited about it is like, uh, so I'm a big believer in, um, multi-agent systems, and so this going back to our, uh, kind of this idea of an atomic agent, uh, so imagine the MCP server, like obviously it calls tools, but the way I think about it, so I'm working on, uh, my current passion project is agent.um, and we'll talk more about the, I think we should, because I think it's interesting, not to promote the project at all, but, uh, there's some interesting ideas in there, one of which is around, we're going to need a mechanism for, uh, if agents are going to collaborate and be able to delegate, um, there's going to need to be some form of discovery, uh, and we're going to need some standard ways, like, okay, well, I just need to know what this thing over here is capable of. We're going to need a registry, which Anthropic is working on; I'm sure others will and have been, uh, doing directories of, of, uh, and there's going to be a standard around that too. How do you build out a, a directory of MCP servers in a standard way? I think that's going to unlock so many things just because, and we're already starting to see it. So I think MCP or something like it is going to be the next major unlock because it allows, uh, systems that don't know about each other, don't need to, just, it's that kind of decoupling of like Sentry and whatever tools someone else was building, and it's not just about, you know, Claw desktop or things like even on the client side. I think we're going to see very interesting consumers of MCP, uh, MCP clients versus just the chat body kind of things like, you know, Cloud desktop and and Cursor and things like that, but yeah, I'm very excited about MCP in that general direction.

I think the typical cynical developer take is like, we have OpenAPI, why what's the new thing? Uh, I don't know if you have a, do you have a quick, uh, MCP versus everything else? Yeah, so it's, so I like open a, right, it's just a descriptive thing; is OpenAPI, OpenAPI? Yes, that's what I meant, uh, is, so it's, it's basically a self-documenting thing; we can do machine gen, lots of things from that output; it's a structured definition of an API. I get that, love it, um, but MCPs, uh, sort of are kind of use-case specific; they're perfect for exactly what we're trying to use them for around LLMs. In terms of discovery, it's like, okay, I don't necessarily need to know kind of all this detail, and so right now we have, we'll talk more about like MCP, uh, server implementations, but we will, I think, I don't know, um, maybe we won't, at least this in my head back process, but I do think MCP as a adds value above OpenAPI, um, it's, yeah, just because solves this particular thing, and if we had come to the world, which we have, like it's like, hey, we already have OpenAPI as like, if that were good enough for the universe, the universe would have adopted it already. There's a reason why MCP is taking off because marginally adds something that was missing before and doesn't go too far, and so that's why the kind of rate of adoption, you you folks have written about this and talk about ITP, yeah, and and it won because the universe decided that this was useful and maybe gets supplanted by something else, and maybe we discover, maybe OpenAPI was good enough the whole time. I doubt that, uh, but we see the, the meta lesson; this is, I mean, he's an investor in Dev tools companies; I work in developer experience at, in Dev tools companies; Y everyone wants to own the standard. Yeah, uh, I'm sure you guys have tried to have launched your own standards, actually, it's HubSpot known for a standard, you know, obviously inbound marketing, but is there a standard or protocol that you ever tried to push?

No, and and there's a reason for this is that, uh, and I don't mean need to mean, uh, speak for the people of HubSpot, uh, but I personally kind of do am not, I'm not smart enough; that's not the, like, I, I think I have a, you're smart enough for that; I'm much better off understanding the standards that are out there, and I'm more on the composability side; let's like take the pieces of technology that exist out there, combine them in creative unique ways, um, and I like to consume standards; I don't like to, and that's not, I like to create them; I just don't think I have the both the raw wattage or the credibility is like, okay, well, who the heck is Daresh, and why should we adopt a standard you created? Yeah, I mean, there are people who don't monetize standards, uh, like Open Telemetry was, is a big standard, and lights that never capitalize on that. So okay, so if I were to do a, there's two things that have been in my head in the past, one around, um, very very basic one around, I even have the domain for, I have a domain for everything, U for open marketing of, cu, the issue we had in HubSpot, grew up in the marketing space, is there was no standard around data formats and things like that; it doesn't go anywhere. But the other one, uh, I did not mean to go here, but I'm going to go here, um, it's called Open Graph; I know the term was already taken, but it hasn't been used for like 15 years now for, um, for its original purpose, but what I think should exist in the world is, right now our information, all of us, uh, nodes are in, uh, the social graph at Meta or the professional graph at LinkedIn, both of which are actually, U, relatively closed and actually annoying ways, um, like very very closed, right, especially LinkedIn, and I personally believe that, um, if it's my data and if I would get utility out of it being open, I should be able to make my data open or publish it in whatever forms that I choose, as long as I have control over it; it's opt-in. So the idea is around Open Graph that says, here's a standard, here's a way to publish it, and I should go to open, I should be able to go to openggraph.org/daresh.json and get it back, and it's like, here's your stuff, right, and I can choose along the way, and people can write to it; I can prove, and there can be entire system, uh, and if I were to do that, I would do it as a, like a public benefit, uh, nonprofit kind of thing, is just as a contribution to society; I wouldn't try to commercialize that.

Have you looked at, at Proto, what's that, at Proto, that's the protocol behind Blue Sky? Okay, my good friend Dan Abramov, who was the, uh, face of React for many many years, uh, now works there, and you know, he, he actually did a talk that I can send you, which basically kind of tries to articulate what you just said, but he does, he loves doing these like really great analogies, which I think you'll like, which is like, you know, a lot of our data is behind a handle, behind a domain, yep, so he's like, all right, what if we flip that? What if it was like our handle and then the domain, so, um, and that's really like, your data should belong to you, Y, and I should not have to wait 30 days for my Twitter data to export, Y, and you should be able to at least be able to automate it or do like, yes, I should be able to plug it into an agentic thing, um, yes, I think we're, cuz so much of our data is, uh, locked up. I think the trick here isn't that that standard, it is getting the normies to care.

Yeah, normies don't care. That's true. But building on that, normies don't care, um, so you know, privacy is a really hot topic and easy word to use, but it's not a binary thing; like there are use cases where, and we make these choices all the time, that I will trade not all privacy, but I will trade some privacy for some productivity gain or some benefit to me that says, oh, I don't care about that particular data being online if it gives me this in return, or I don't mind sharing this information with this, uh, company if I'm getting, you know, this in return, but actually sort of should be my option. I think now with computer use, you can actually automate some of the exports; like something we've been doing, um, internally, it's like everybody explores the LinkedIn connections, and then internally we kind of merge them together to see how we can connect our companies to customers or things like that, um, and not to pick on LinkedIn, but, uh, since we're talking about it, but they are feel strongly enough on the, you know, do not take LinkedIn data, they will block even browser use kind of things or whatever, they go to great, great lengths, uh, even to see patterns of usage and it says, oh, there's no way you could have, you know, gotten that particular thing or whatever, without, and it's, um, so it's, there's wasn't there a Supreme Court case that they lost? Yeah, so the one they lost was around someone that was scraping public data that was on, uh, on the public internet, and that particular company had not signed any terms of service or whatever; it's like, oh, I'm just taking data that's on, there was no, uh, and so that's, that's why they won, but now, you know, the question is around, can LinkedIn, and I think they can, like when you use, as a user, you use LinkedIn, you are signing up for their terms of service, and if they say, well, this kind of use of your LinkedIn account that violates our terms of service, they can shut your account down, right? They can, uh, and they, yeah, so anyway, we don't want, need to make this discussion, by the way, I, I love the company, don't get me wrong; I'm an avid user of the product, uh, you got, yeah, I mean, you got over a million followers on LinkedIn, I do, and know people there for a long, long time, right, that I have lots of respect, and I understand even where the mindset originally came from of this kind of Members First approach to, uh, you know, privacy first; I, I sort of get that, but sometimes you sort of have to wonder, it's like, okay, well, that was 15, 20 years ago; there's likely some controlled ways to expose some data on some members' behalf and not just completely be a binary; he like, no, thou shalt not have the data, um, I just pay for Sales Navigator, so, um, right before we move to the next layer of abstraction, anything else on MCP? You mentioned, um, let's move back, and then I'll, I'll tie it back to MCPs, um, so I think the open this with agent.

Okay, so I'll start with, here's my kind of running thesis is that as AI and agents evolve, um, which they're doing very, very quickly, we're going to look at them, uh, more and more, I don't like to anthropomorphize; we'll talk about why this is not that less as just like raw tools, uh, and more like teammates; there'll still be software; they should self-disclose as being software; I'm totally cool with that, but I think what's going to happen is that, uh, in the same way you might collaborate with a team member on Slack or Teams or whatever you use, you can imagine a series of agents that do specific things, just like a team member might do that you can delegate things; you can collaborate; you can say, hey, can you take a look at this, uh, can you proofread that, can you try this, can you whatever it happens to be. So I think it is, I will go so far as to say it's inevitable that we're going to have hybrid teams someday, and what I mean by hybrid teams, so back, back in the day, hybrid teams were, oh, well, you have some full-time employees and some contractors; then it was like, hybrid teams are some people that are in the office and some that are remote; that's the kind of form of hybrid; the next form of hybrid is like the carbon-based life forms and agents and AI and some form of, form of software. So let's say we temporarily stipulate that I'm right about that, uh, over some time horizon, that eventually we're going to have these, U, kind of digitally hybrid, uh, teams. So if that's true, then the question you sort of ask yourself is that then what need to exist in order for us to get the full value of that new model? It's like, okay, well, you sort of need to like, okay, well, how do I, if I'm building a digital team, like how do I, just in the same way if I'm interviewing, uh, for an engineer or a designer or a PM, whatever, it's like, well, that's why we have professional networks, right? It's like, oh, they have a presence on, likely LinkedIn; I can go through that semi-structured, structured form, um, and I can see the experience or whatever, uh, you know, self-disclosed, but okay, well, agents are going to need that someday, uh, so I'm like, okay, well, this seems like a thread that's, uh, worth pulling on that says, okay, so I create, so agent.a, um, is, is out there, LinkedIn for agents; it's LinkedIn for agents; it's a professional network for agents, and the more I pull on that thread, it's like, okay, well, if that's true, like what happens, right? It's like, oh, well, they have a profile, just like anyone else, just like a human would; it's going to be a graph underneath, just like a professional network would be; it's just that, and you can have, it's, uh, you know, connections and follows, and agents should be able to post; that's maybe how they do release notes, like, oh, I have this new version, whatever they decide to post, um, it should just be able to behave as a node on the network of a professional network. As it turns out, the more I think about that and pull on that thread, the more and more things like start to make sense to me, so it may be more than a just a pure professional network, so, um, so my original thought was, okay, well, it's a professional network, and agents as they exist out there, which I think there's going to be more and more of, will kind of exist on this network, um, and how the profile, but then, and this is always dangerous, I'm like, okay, I want to see a world where thousands of agents are out there in order for the, because those digital employees, the digital workers, don't exist yet, uh, in any meaningful way, and so them like, oh, can I make that easier, uh, for like, and so I have, as one does, it's like, I'll build a low-code platform for building agents; how hard could that be, right? Like very hard, as it turns out, U, but, uh, it's been fun. So now agent.a has 1.3 million users; 3,000 people have actually, you know, built some variation of an agent, sometimes, sometimes just for their own personal productivity, uh, about a thousand of which have been published, uh, and the reason this comes back to MCP for me, so imagine, uh, that in other networks, but just, uh, since I know agent.a, so right now, uh, we have an MCP server for agent.a, uh, that exposes all the internally built agents that we have that do like super useful things, like, you know, I have access to a Twitter API that like I can subsidize the cost, and I can say, you know, if you're looking to build something for social media, these kinds of things with a single API key, and it's all completely free right now; I'm funding it, um, that's a useful way for a developer to say, oh, I have this idea; I don't have to worry about OpenAPI; I don't have to worry about, now, you know, this particular model is better, has access to all the models with one key, and we proxy it kind of behind the scenes, um, and then expose it, so then we get this kind of community effect, right, that says, oh, well, someone else may have built an agent to do X; like I have an agent right now that I built for myself to do domain valuation for website domains because I'm obsessed with domains, right? And like there's no efficient market for domains; there's no Zillow for domains right now that tells you, oh, here, what houses in your neighborhood sold for; it's like, well, why doesn't that exist? We should be able to solve that problem, and yes, you're still guessing, fine, uh, there should be some simple heuristic, so I built that; it's like, okay, well, let me go look for past transactions; you say, okay, I'm going to type in agent.a agent.com, whatever domain, uh, what's it actually worth? I'm looking at buying it; it can go and say, oh, and which is what it does; it's like, I'm going to go look at, are there any published, uh, domain transactions recently that are similar? I use the same word, same top-level domain, whatever it is, and it comes back with an approximate value, and it comes back with its kind of rationale for why it picked the value and comparable transactions; oh, by the way, this domain sold for, published. Okay, so that agent, now, let's say existed on the web, uh, on agent.a, then imagine someone else says, oh, you know, I want to build a brand-building agent, uh, for startups and entrepreneurs to come up with names for their startup, like a common problem; every startup's like, I don't know what to call it, and so they type in five random words that kind of define whatever their startup is, uh, and you can do all manner of things, uh, one of which is like, oh, well, I need to find the domain for it, uh, what are possible choices? Now it's like, okay, well, it would be nice to know if there's an aftermarket price for, if is it listed for sale? Awesome. Then imagine it calling this valuation agent; it's like, okay, well, I want to find where the arbitrage is; where the agent valuation tool says this thing is worth $225,000; it's listed on GoDaddy for $55,000; it's close enough; let's go do that, right? And that's a kind of composition use case that in my future state, thousands of agents on the network, all discoverable, something like MCP, and then you as a developer of agents have access to all these kind of Lego building blocks based on what you're trying to solve, um, then you blend in orchestration, which is getting better and better with the reasoning models. Now just describe the problem that you have. Now the next layer that we're all contending with is that how many tools can you actually give an LLM before the LLM breaks? That number used to be like 15 or 20 before, um, you're kind of man, start of very dramatically, and so that's the thing I'm thinking about now; it's like, okay, if I want to expose a thousand of these agents to a given LLM, obviously I can't give it all thousand; is there some intermediate layer that says, based on your prompt, I'm going to make a best guess that which agents might be able to be, uh, helpful for this particular thing?

Yeah, yeah, like RAG for tools, um, I did build a latent-based researcher on agent.a. Okay, nice. Yeah, that seems like a, you know, then there's going to be a latent-based scheduler, and then once I schedule research, you know, and you build all of these things, by the way, my apologies for the user experience, uh, you realize it, it's pretty good; I, a normy-friendly thing; that's your magic; HubSpot does the same thing. Yeah, just to like quickly run through it, you can basically create all these different steps, and these steps are like, you know, static versus like variable-driven things. How did you decide between this kind of like low-codish versus doing, you know, low code with code back end versus like not exposing that at all? Any fun design decisions? Yeah, and this is, I think, um, I think lots of people are likely sitting in exactly my position right now, now coming through the choosing between deterministic, like if you're like in a business or building, you know, some sort of agentic thing, you decide to do a deterministic thing, uh, or do you go non-deterministic and just let the

Alum handled it right, um, with the reasoning models, the original idea, and the reason I took the low-code, stepwise, a very deterministic approach, is the um reasoning models did not exist at that time. That's thing number one. Thing number two is, if you can get—if you know in your head what the actual steps are to accomplish whatever goal—why would you leave that to chance? There's no—there's literally no upside. Just—just tell me like what steps do you need executed.

So right now what I'm playing with, um, so one thing we haven't talked about yet, and and people don't talk about UI and agents, uh, right now the primary interaction model, uh, or they don't talk enough about it. I know uh some people have, but it's like, okay, so we're used to the chatbot back and forth. Fine, I get that. I think we're going to move to a blend of—some of those things are going to be uh synchronous as they are now, but some are going to be async. It's just going to put in a queue, just like—and this goes back to my—man, I talk fast—um, but um I have this—I only have one other speed; it's even faster—um, so imagine it's like if you're working—so back to my—oh, this—we're going to have these hybrid digital teams. Like you would not go to a coworker and say, "I'm going to ask you to do this thing," and then sit there and wait for them to go do it. Like that's not how the world works. So it's nice to be able to just like hand something off to someone. It's like, "Okay, well maybe I expect a response in an hour or a day or something like that," and there's some implicit contract that we have um with our co-workers in terms of when things uh need to happen.

So the UI around agents—so if you look at the output of agent. Agents right now, they are the simplest possible manifestation of a UI, right? That says, "Oh, we have inputs of like four different types," like we've got a drop-down, we've got multi-select, all the things. It's like back in HTML, uh, the original HTML 1.0 days, right? Like you're the smallest possible set of primitives for UI, uh, and it says, "Okay, because we need to collect uh some information uh from the user," and then we go do steps and do things and generate some output in HTML or markup are the the two uh primary examples.

So the thing I've been asking myself—if I keep going down that path—like, so people ask me, I get requests all the time, is like, "Oh, can you make the UI?" It's sort of boring. I need to be able to do this, right? And if I keep pulling on that, it's like, okay, well now I've built an entire UI Builder thing. Where does this—like where does this end? And so I think the right answer—and this is what I'm uh going to be back-coding once we I get done here—is around injecting a code generation, UI generation into the agent. flow. Right? As a builder, you're like, "Okay, I'm going to describe the thing that I want," much like you would do in a Vibe coding world, but instead of generating the entire app, it's going to generate the UI that exists at some point uh in either that deterministic flow or something like that says, "Oh, here's the thing I'm trying to do. Go generate the UI for me," and I can go through some iterations um and what I think of it as a—um, so like I'm going to generate the code, generate the code, tweak it, go through this kind of prompt style like we do with Vibe coding now, and at some point I'm going to be happy with it, and I'm going to hit save, and that's going to become the action in that particular step. It's like a caching of the generated code that I can then—not like—inference time cost is just a actual code at that point.

So yeah, I invested in a company called e2b, which does code sandbox, and they powered the LM Arena web Arena. So it's basically the—just like you do LMcs, like text to text, uh, they do the same for like UI generation. So if you're skinning a model, how do you do it? Um, but yeah, I think that's kind of where—that's the thing I'm really fascinated by.

So the early um LLMs, you know, were understandably but laughably bad at simple arithmetic, right? That's the thing—like my wife—normies would ask us, like, "You call this AI? Like it can't—" My son'd be like, "It's just stupid. It can't even do like simple arithmetic." And then like we've discovered over time that—and there's a reason for this, right? It's like it's a large—there, you know, the word "language" is in there for a reason in terms of what it's been trained on. It's not meant to do math. But now it's like, okay, well the fact that it has access to a Python interpreter that I can actually call it runtime that solves an entire body of problems that it wasn't trained to do, and it's basically a form of delegation. And so the thought that's kind of rattling around in my head is that that's great. So it's—it's like took the arithmetic problem, took it as far as now—like anything that's solvable through a relatively concrete um Python program, it's able to do a bunch of things that I couldn't do before. Can we get to the same place with UI? I don't know what the future of UI looks like in an agentic AI world, but maybe let the LLM handle it, but not in the classic sense. Maybe it generates it on the fly, or maybe we go through some iterations and hit cache or something like that, so it's a little bit more predictable. Uh, I don't know, but yeah, and especially when it's the—supposed to intervene—so especially if you're composing them, most of them should not have a UI, because then they're just web hooks to to somewhere else.

I just want to touch back—I don't know if you have more comments on this. I was just going to ask, when you—you said you—you're going to go back to code—what do you coding with? What's your stack? Yeah, uh, so Python's my language. Uh, I'm glad that it won in terms of the AI languages lingua franca, the second best language for everything, by the way. There—I think exactly n of one of things that I disagree with Brett Taylor on uh when he was on—and just generally I'm a massive Brett Taylor fan—smart, one of my favorite people in Tech. Like it was like a segment in there he was talking about like, "Oh, we need a a different language than Python," or whatever, that is like built for uh built for AI and built—it's like, "No, Brad, I don't think we do actually. It's just fine." Um, it deals with—just fine, just expressive enough, and and it's nice to have a language that we can use as a common denominator across both humans and AI. It's—it doesn't slow the AI down enough, but it does make it awfully useful for us to also be able to participate in that kind of future world—that we can still be somewhat useful, anyway. But yeah, so it's Python, uh, cursor as my uh kind of code gen thing.

Yeah, I would also mention that I really like your code generation thing. I have another thesis I haven't written up yet about how generative UI has kind of not fulfilled its full potential. We've seen the Bolt and Lovvables, and those are great, and then Vercel has a version of generative UI that is basically function calling pre-made components, and there's something in between where you should be able to generate the UI that you want and pin it and stick to it, and that becomes your form—your—yeah—and and so the way I put it is—um, you know, I think that the two form factors of agents that have seen a lot of product-market fit recently has been deep research and the AI Builders, like the B-Lovvables. I think there's some version of this where you generate the UI, but you—you sort of generate the Mad Libs, fill-in-the-blanks forms, and then you—you—you keep that stable Y, and the deep research just fills that in. Yeah. Yep, and that's it. I like that. Yeah. Um, so I—I—I love those uh kind of simple—simple implementations and kind of abstractions, but then if you look at the—I'll say almost like the polar opposite of that—so so right now most of the UIs that you and I think about or conceive or even examples are based on the primitives and the vocabulary that we have for UI right now. It's like, oh, we have text boxes, we have check boxes, we have radio buttons, we have pull-downs, we have nav, we have clicks, touches, swipes, now voice, whatever it is—the set of primitives that exist right now. We will combine them uh in interesting ways, but where AI is going to be headed on—I think on the UI front—is the same place as headed on the science front—that originally it's like, oh, well based on the things that we know right now, it'll sort of combine them, but we're like right at the cusp of it being able to—actual novel research. So maybe a future version of AI comes up with a new set of primitives that actually work better for human-computer interaction than things that we've done in the past, right? It's like I don't think it's it ended with the uh the checkbox, radio button, and drop-down list, right? I think there's life beyond that. Uh, yeah, I know we're going to move to business models after, but when you talked about hybrid teams, one way we talk to folks about is like you had offshoring, you had onshoring, which is like, you know, moved to a cheaper place in the country than offshore, and now it's like AI shoring—you're kind of moving somewhere else to a—that's the thing people say—AI shoring. Yeah, that's the first time I've ever heard of that. Yeah, I don't know, man, but I—I think to me the most interesting thing about the professional networks is like with people, you have limited availability to evaluate a person, so you have to use previous signal as kind of like a evaluation thing. With agents, theoretically you can have kind of like proof of work, you know, you can run simulations and like evaluate them in that way. How do you think about that when running—building agent. even, you know, instead of just choosing one, I could like literally just run across all of them and figure out which one is going to work best?

I'm a big believer—so uh under the covers when you build an—because the primitives are so simple, you have some set of inputs—we know that what the variables are—every agent that's on agent.a automatically has a REST API that's callable in exactly the way you would—you would expect—automatically shows up in the MC—MCP server, so you're able to invoke it in whatever form you uh decide to. And so my expectation is that in this future state, whether it's a human hiring uh agent to do a particular task or evaluating a set of five agents to do a particular task and picking the best one for their particular use case, we should be able to automate that. It's like, "I just want to try it," um, and there should be a policy that the publisher, builder of the agent has that says, "Okay, well I'm going to let you call me 50 times, 100 times before you have to pay," or something like that. Uh, we should have effectively like an audit trail, like, "Okay, this agent has been called this many times." We also have uh kind of human ratings and reviews right now, and we have tens of thousands of reviews of the existing agents on agent.averages like 4.1 out of five stars, and all those things are nice signals to be able to have, but the the kind of callable uh verifiable kind of thing I think is super useful. Like if I can just call an—give me an API that says, "Here are five agents and it solves this particular problem for me," if I have like a simple eval, I think that'd be so powerful. I wish I had that for humans, honestly. That would be so cool. Yeah, because I mean when I was running engineering teams, people would try and come out with these rubrics, you know, when hiring, and it's like they're not really helpful, but you just kind of—some ground truth. But I feel like now—say you want to hire—yeah, an AI software engineer—you can literally generate like 15, 20 examples of like your actual issues in your organization, both from a people perspective of like collaboration and like actual code generation, and just pay for it to run it. Like today we do take-home projects and we pay people—this should be kind of the same thing. It's like, "I'll just run you," but I feel like people are not investing in their own eval as much—that's present company included, right? Uh, like everyone talks about eval, everyone accepts the fact that we should be doing more with evals—I won't say nobody, but almost nobody actually does—that's the—and yeah, and it's a topic for a whole other day. I'm—it's funny, I mean, because obviously hpot is famous for launching graders of things. Yes, you'd be perfect for it. So—I know—I agreed on eval, by the way. Uh, I mean, I just force myself to be the human in the loop, or you know, someone I work with, and and that's okay, but obviously the scalable thing needs to be done.

Just fun fact on—or question on AI agent AI—you famously—you've already talked about the chat.com acquisition and all that, and that was around the time of custom GPTs in the GPT store launching, and I definitely feel agent AI is a kind of the GPT store but not taken seriously. Why do you feel OpenAI, if like they woke up one day and they were like, "Agent AI is the thing. Like we should just reinvest in GPT store"—is that a fear?

Um, I think that's an—is not—won't be. Agent. uh, driven. It'll—it's an inevitability that OpenAI—I don't have any insider information. I'm an investor, but no insider information, is because it makes too much sense, um, but for them not to—like, and they—they've taken multiple passes at it, right? They did the plugins back in the day, then the custom GPTs, and then the GPT store, because, you know, being the platform that they are, I think it's inevitable um that they will ultimately come up with—and they already have customers—it's going to happen. I'm not—on the list of things I promised myself I would never do is compete with Sam Altman ever—not intentionally, anyway—um, but here you are—but yeah, here I am. But I'm not really—right, it's—um, not really. It's—UI—I mean, it's free, so like whatever, but you know, at some point—but I mean, actually valuable—he's—they're solving a much, much bigger problem. I'm like a small, tiny rounding error in the universe—um, but the reason that compelled me to actually create it in the first place, because I knew custom GPTs existed, and I did have this uh rule in my head that you don't compete with Sam—he's literally like at the top of my list of people not to compete with—he's so good—um, but the thing that I needed in terms of—for my own personal use—which is how agent. got started, because I was building a bunch of what I call solo software things for my own personal productivity gain, and I found myself doing more and more kind of LLM-driven stuff because it was better that way—you know, AI sort of showed up in those uh solo projects a bunch—and so the thing I needed was an underlying framework to kind of build these things, and high on the list was, "I want to be able to uh straddle models," because certain steps in the thing is like, oh, for this particular uh thing uh involves writing some—maybe I want to use cloud for this particular thing, maybe I want to do this, even around image generation, different types of uh whether it has text or doesn't have text, whatever—and I want to be able to mix and match, and my sense is that whether it's OpenAI or Anthropic or whatever, they're likely going to have an affinity for their own models, right? Which makes sense for them, but I can sort of be—for my own purposes and for our user base—a little bit of the Switzerland. It's like we don't think there's like one model to rule them all. Based on your use case, you're going to want to mix and match and maybe even change them out—maybe even test them—back to the kind of eval idea—it's like, "I have this agentic workflow," and here's the thing we've been playing with recently because we have enough users now where they like the LLM and I look at the bills uh and it's like, "Oh, I'm spending real money now," um, and and this is just human nature, right? It's not just uh normies, but it's like—so you have this, you know, drop-down of all the models um that you can say which model you want to use in your agent.a agent, and as it turns out, people pick the largest number, so they will pick GPT 4.5 or whatever—whatever it is—right? It's like, "It's—oh my god—GPT 4.5—yes—ouch—yes"—yeah—but the thing I've promised myself um is we will support all of them, regardless of what it costs. Like, once again, I see this as a just a research thing, uh, you know, benefit to humanity, and inference costs are going down—at least I—so I tell myself late at night to uh so I can sleep—um, so they picked the uh the highest numbered one, and so we have an option in there right now that says—and which is the first option—it's like, "Let the system pick for me—auto." As it turns out, people don't do that. They just pick the GPT because they don't trust it yet, which is fine—they shouldn't trust it uh completely—but one thing we discovered is that um if we back-channel it—and this is the thing we're testing with—is that, oh, if I can just run the exact same agent uh that gets run a thousand times, we'll do it on our own internal agents first, and if the ratings and reviews—because we're getting human evals all the time on these agents—we can get a dramatic multiple—orders of magnitude reduction by going to a lower model with literally like no change in the quality of the output, right? Which makes sense, because so many of the things we're doing doesn't require the most powerful model, um, and it's actually B because there is higher latency—it's not just a cost thing—but um, so anyway, like in that kind of future state, I think we're going to have model routing and uh whole body of people working on that problem, too. It's like uh help me pick the best model at runtime.

Would you buy or build model routing? I buy everything that I can buy. I—I—I don't want—I don't—I don't want to build anything uh if I don't have to. One of the most impressive examples of this I think was our Chai AI conversation, which I think about a lot. He views himself explicitly as a marketplace. You are kind of a marketplace, but he has a third angle, which is the model providers, and he lets them compete, and I think that sort of three-way marketplace maybe makes a lot of sense. Like I don't know why every AI company isn't built that way.

It's a good point, actually. Yeah, um, it makes sense. I—I'm a—list of things I'm super passionate about—I'm very passionate about uh efficient markets or—and or extremely irritated by inefficient markets—and so efficient markets—for the normies listening—are markets that exist where every possible—efficient markets are the ones that every transaction that should occur actually does. That's an efficient market—that that should happen. And so then why do inefficient markets exist? Well, maybe the buyer and seller don't know about each other, uh, maybe there's not enough a trust mechanism, there's no way to actually price it or come up with fair market value, fair pricing. And as you kind of knock those dominoes down, you know, the market becomes more and more—and lots of uh latent value exists as a result of inefficiency, and whoever removes those inefficiencies for like high-value markets makes a lot of money. That's been proven time and time again. This is one of those examples of—there's an inefficiency right now because we are like either over—using over-models or whatever—let's just reduce that to an efficient Mar—the the right model should be mashed up with the right use case for the right price, uh, and then we'll—very, very interesting.

You ever looked into DSP? I have looked at it—not deeply enough, though. It's supposed to be—as far as I think—the the the only like evals-first framework, right? And if eval is so important—and by the way, the the the relationship between this and all that is DSP would also help you optimize your models—yep—because you did the eval first—yep—I wonder why it's not—not as popular, you know—um, but I mean it is growing—interaction—I would say we're—we're keeping an eye on it.

Let's talk about business models. Obviously, you have kind of two—work as a service and results as a service. Why—I'm curious how you divide the the two. Yeah, uh, so work as a service is—so we—we know about software as a service, right? So I'm licensing software that's delivered to me as a service. That's been around for decades now, um, so we understand that, but the consumer of that service is generally a human uh that's doing the actual work uh whichever software you're buying. Work as a service is the software is actually doing the work, whatever that work happens to be. And so that's work as a service—so uh come up with kind of discrete use cases, whether it's kind of classification or legal contract review or whatever—the software is actually doing the thing. Uh, results as a service is you're actually charging for the outcome, not actually the work, right? That says, "Okay, instead of saying I'm going to pay you x amount of dollars to review a legal contract," or this amount of time or number of uses or something like that, "I'm going to actually pay you for the actual result," um, which is—so my take on this—and in the industry or parts of the industry are super excited about this kind of results service or outcomes-based pricing—and I think the reason for—I think we're over-indexing on it—and the reason we're over-indexing on it is the most popular use case on the kind of agent side right now is like customer support. Well-documented, a lot of the ERS that have U—you know, agents uh for customer support do it on a number of tickets resolved—times, you know, X doll—you know, per ticket—and the reason that that makes a lot of sense is that the customer support departments and teams sort of already have a sense for what a ticket uh costs to resolve through their kind of current—current way—and so you can come up with an approximation um for a what the kind of economic value is. There's also a at least uh semiobjective measure for what an acceptable uh resolution or outcome is, right? Like you can say, "Oh, well we measured the net promoter score or cat uh for tickets," or whatever—"as long as the customers—90% of the tickets were handled away—the customer was happy," that's whatever your kind of line is—as long as the AI is able to kind of replicate that same SLA—it's like, "Okay, well it's the same—they're fungible—one versus the other." I think the reason we're over-indexed, though, is that there are not that many use cases that have those two dimensions to them that are objectively measurable and that there's a known economic value that's constant—like customer support tickets, because they're handled by humans, makes sense, and humans have a kind of discrete cost, and especially in retail, which is where this originally got started—in B2C companies that have a high volume of customer support tickets that are distributing across a ticket is roughly worth the same because it takes the same amount of time for most humans to do that kind of level-one uh tier-one support, but in other things, the value per uh outcome can vary dramatically—literally by orders of magnitude—in terms of what the thing is actually worth. That's kind of thing number one. Thing number two is, how do you objectively measure—so let's say you're going to do a uh a logo creator as a service based on results, right? And that's a completely opposite—subjective thing or whatever—and so okay, well it may take me 100 iterations—it take me five iterations—the quality of the output is actually not completely under my control—it's not up to software—it could be—you have weird taste or you describe what you're looking for enough or whatever—it's like it was just not a solvable problem. Design—kind of qualitative, subjective disciplines deal with this all the time—how do you make for a happy customer? There's a reason why they have—"Oh, we'll go through five iterations," but you know, "Our output is—we're going to charge you $5,000 or $500," whatever it is, "for this logo," but that's hard, right? To kind of do it scale.

So just a a relatable anecdote—we—our podcast—actually, we just uh got a new logo, and we did a 99 Designs for it, and there's so many designers who are working really hard, but like I just didn't know what I wanted. Yeah, so they were—I was like, "Just too bad. Like I know—like you—you seem great, but you know—" Yep. Yeah, that's another example of a a market made efficient, right? It's like I've been a 99 Designs user and customer for dozen-plus years. It's fantastic—so many designers—like this doesn't cost that much for them to do—it's worth a lot to us—we can't design for [ __ ]—um—yeah. Yep. By—pro tip on 99 Designs—um, is that on the margin you're better off kind of committing to paying the designer that you're going to pick a winner, whether you like it or not—doesn't really matter—uh, and that gets higher participation, which—and the—you're still going to get a bunch of crap that happens—you—a bunch of noise in it—uh, but the kind of quality outcome is often a function of the number of iterations—uh, and logo design is one of those examples. If you can—if you had to choose between 200 logos versus 20 logos, chances are closer that you're going to find something you like. Yeah, uh, for those—I have a blog post on my reflections on the 99 Design thing, and that's on—that's one of those—that they—they give an estimate of uh how many designs you get, and I think that the modifier for like—"We will pay you—we'll pay somebody," and maybe it's you—"is like 30 to 60," but actually it's 200. Yep, so it's underpriced. Yep. Yep.

Do you think some markets are just fundamentally going to move to more results-driven business models? Probably. And I don't understand enough markets well enough—no—but if we had to kind of sort order—rank them—there's likely some dimension along which we could sort that. It's like—how these kinds of businesses—is there an objective measure of uh of kind of truth or uh the outcome—um—is there a way to kind of price it—um—in terms of—so low variance or variability on uh on the value per outcome? If those things are true—whatever industries that is true in—customer support uh is an example—there's likely lots of other examples where uh those two things are true, but then the thing I wonder, though, is that from the customer's perspective, would they rather actually pay uh for work as a service versus an actual—it's like maybe the way they think about it is that's sort of my arbitrage opportunity—like in the—I can work done for X, but the value is actually Y—why would I want that Delta to be squo out by the kind of provider of the software if I have a choice? I don't know. Oh, I mean, okay—attribution—like there's, you know—

There's 18 things that go into that; you're one of them. So, like, you know, it's it's hard to tell.

Yes, it is. So yeah. By the way, have you seen—obviously you're in this industry, not exactly HubSpot's exact part of the market—but what have you seen in attribution that is interesting? You know, if because that that directly ties into work as a service versus results.

Yeah, not enough, because we are so, um, as a world, as an IND—just pick your thing—so behind on track. This is why I think Web3, uh, in the way that it was meant to be done, yeah, is going to make a comeback, because fundamental principles of that make sense. I think what happened in that world was, got a bunch of Crypt Bros and grifters and NFT stuff or whatever, that was loosely related. There was no actual—but the idea of a, of a blockchain, of a trackable thing, of, you know, being able to fractionalize digital assets, attribution, having an audit log, a published thing that's verifiable—all those primitives make sense, right? Like, and maybe there's, you know, a limited—but it's not zero—set of use cases where the kind of what we would now call like the inference cost or the the overhead, the tax for storing data on the blockchain, it has a—and there's certainly a tax to it—it doesn't make sense for all things, but it makes sense for some things, uh, for sure. So, but we just don't have like attribution in any meaningful way. I, I don't think.

Isn't it sad that it's so important, and I know no answer? It's partly comes down to incentives, uh, so the people that actually have the data or parts of the data from which attribution could be calculated or derived don't really have the incentives to make that data available. Um, so even something as simple like on, uh, like the PPC side, right, on the Google search thing, uh, which, you know, that's sort of my world or has been, we have less data now than we did back in the day in terms of like click-throughs and things like that. Before Google would actually send you, "Here are the keywords people typed," and you know, years ago they, you know, they even took that away. So it's hard to kind of really connect the dots back on things, and we're seeing that across—it's not just PPC, but just—they took that away from Search Console.

What's that? Their Search Console has that? They took that away? Search Console has that, but your website, if you go to Google Analytics, you can connect it back to the Google Search Console.

I see. I see. Yeah. Yes. Uh, okay. All right. Yeah. Well, it's a known thing; you don't have to make it a r about Google.

What about software engineering? Do you think it will stay as like a work as a service, or do you think—I think most companies hire a lot of engineers but didn't really know what to do with them, or like didn't really use them productively. Yeah, and I think now they're kind of hitting this like, you know, crisis where it's like, okay, I don't know what I would price an agent because I don't really know what my people are doing anyway. Like, how do you think that changes?

I think, um, so I'm actually bullish on engineers in terms of their kind of long-term economic value, um, not despite all the movements in COIN and all the things that we're, you know, already seeing, but because of it, uh, because what's going to happen as a result of AI, and people have talked about this, um, in, um, even other disciplines, we're going to be able to solve many more problems. Semi-math guy me is like, okay, so we always say, oh, well, now you agents are going to be doing code or whatever, and so there's going to be a million software engineer, you know, virtual digital software engineers out there, and so the value per engineer is going to go down because I'm just in that that same mix. I, as an engineer, what they don't recognize is that it's not just about the denominator; there's a numerator as well, which is what's a total economic value that's possible, and I would argue that's growing faster than the kind of denominator is, that the actual economic value that's possible as a result of software and what engineers, uh, can produce, you know, with the tools that they will have at hand. Um, so I think the value of an engineer actually goes up; they're going to have the power tools; they're going to be able to solve a larger base of problems that are going to need to be solved.

Yeah, it feels to me like he'll stay as like work as a service, pay work. I don't think there's like a way—there will be a set of, um, engineers that—and we see this all the time, you know, there, uh, in the media industry, you have people that are kind of writers, but then you have freelancers that, you know, write articles or write however they manifest their kind of creative talent, and both make sense, right? There's like the work for hire; there's also the kind of outcome-based, or like, "I produce this thing," and maybe they—some of those engineers actually produce agents, so they put it on a marketplace like Agent DeSomay, and that's how they make their millions.

Yeah. Any other thoughts just on agents? We got a lot of like M that we want to talk to you about. I think we covered a lot of territory, so I'm, uh, excited about agents. My kind of message to the world would be, "Don't be scared." I know it's scary—easy for me to say as a techno-optimist—but learn it, even if you're a normie, even if you're not an engineer, if you're not an AI person—you'll think of yourself as an AI person—use the tools. I don't care what role you have right now, where you are in the workforce, uh, it will be useful to you, um, and start to get, get to know agents, use them, build them. And I think my message for engineers is always like, there's more to grow; like, we're still early in the early days of figuring out what an agent's, uh, stack looks like, and, uh, I want to push people towards agents with memory.

Yeah. All right. Agents with planning. Oh, we have to talk about memory; we got to talk about memory. Yeah, let's do it, because I think that's the, uh, that's the next—in my mind—the next frontier is actual long-term memory, both for agents, uh, and then for agentic networks in a trustable, verifiable—I won't say privacy first, but, uh, privacy-oriented way. I have an issue with the, the term "privacy first," um, because a lot of times we say "privacy first" when we don't really mean that. Like, "privacy first" means, "I value that above all things; doesn't matter what we're talking about," and that's just not true, not for any, for any human, anything that wants to be used. Um, so it's—but, uh, but so memory is an interesting thing, right? So the thing I'm working on right, right now—lots of things in play in agent.—is around implementation, uh, of memory, and there are great projects out there—mZero being one of them—but the thing that's interesting for me right is—and so we see this in, uh, ChatGPT, other things right now where it does have the notion of a longer-term memory, can pull things back into, into context, um, as needed. The thing I'm fascinated by is, uh, cross-agent memory. So if I'm an agent, uh, builder right now, it's like, okay, here are the things that I sort of know or I, I learn from the user, um, in terms of pulling out the—I'll call them knowledge nuggets for lack of a better term—and that's great, but then when the next agent builder comes out and it's the same user, shouldn't all the things that agent one learned about me, if it's going to be useful for agent two? And as long as I opt into it, it's like, yeah, I don't care—those things—in fact, I would find it awfully annoying to tell agent two and agent n and agent n plus one all the same things I've already told it, uh, because it, it should know—like the system should know—and this is part of the reason why I'm like a believer in these kind of networks of agents and shared state is that that user utility gets created as a result of having shared, shared memory. Not just, we should solve the memory problem for an independent agent, but then we should also be able to share, um, share that context, share that memory across agents, and that's part of the value prop for agent. is like, okay, when you're building, it's like—so we've got, you know, and, uh, one, you—whatever—million users, and we're going to have growing memory about all them. So instead of you going off on your own thing and building an agent out as this kind of, uh, disconnected node in the, in the universe or whatever, here's the value for building on, on the network, on the platform—ours, ours or someone else's—because more—there's more user value that gets created.

It's volatility. How do you think about OAuth for that? Because part of memory is like selective memory. So take like scheduling—I want you to have access—if I have an scheduling agent, you should be able to access the events you're a part of and like what times I have available, but it shouldn't tell you about other events on my calendar. Like, what's that like?

I have so many thoughts on this. This is—and the like the opportunity out there, like solving these kind of fundamental—like this is going to need to exist, right? So right now the closest approximation we have is is OAuth, right? And everyone has—it's like, okay, approve, and it's a very, very, uh, set of scopes, right? Like, based on the, the provider of the, the OAuth server—be Google, whoever it is—HubSpot doesn't matter; it's like, oh, I pick a set of scopes, and they could have defined the scopes to be super granular, uh, but it's sort of up to them, but that is going to move so slowly, right? So for instance, the use case I have right now, like I use email for everything; I use it as a, um, like an event and data for my life, right? And why I mean that like literally, it's like I'm like, anything that I do, if there's a way to kind of get that into email—CU—I know it's an open protocol, right?—like, okay, I will be able to get to that data in useful ways. This is before—so I have 3 million that I've built a vector store off of to solve my own, uh, personal use cases. So I'll give you the example, but obviously I'm not going to build my—all my own software for everything, but if a startup comes along and says, "DMes, can you make your email inbox available in exchange for these things?" I'm like, "Hell no!" Like, that's the literally my crown—like everything—like my life is in here, right?

So you need to share subsets. Yes. And so I think there's a—and maybe this is not the actual implementation—but imagine if someone said, okay, I have a trusted intermediary for that, for trust however defined, that says, okay, I'm going to OAuth to this thing and it gets to control—I can say in natural language—"I only want to pass email to this provider, uh, where the label is one of X, or that's within the last thing, and no more than 50 emails in a day," or whatever. So I don't have them dumping the entire 3 million, uh, you know, back—whatever controls I want to put on it. It's unlikely that the all the OAuth, um, server sizes right now—the Googles, even the big ones, small ones, doesn't really matter—are going to do that, but this is an opportunity for someone, and they're going to need to get to some scale, build some level of trust that says, okay, I'm going to hand over the keys to this intermediary. Yeah, but then, uh, it opens up a bunch of utility because it gives me control, uh, more fine, fine-grain control.

Yeah. I say LangChain has, has an interesting one. There are a bunch of people who has tried to crack AI email. Every single one of them who's tried has pivoted away. Yep. And I'm waiting for Superhuman to do it. Why? I don't know why they haven't, but, you know, some point—that's some cool AI stuff, but they need—I think the pace is to increase, but I think this goes back to like Open Graph, yeah, right? Which is like, I think Google is not incentivized to build better scopes. Nope. And like, they're just not going to do it. No. So we can't even get like—we haven't been able to get semantic search out of Google for like—still totally—yeah, just now they made the announcement this week.

What do you mean, semantic search in Gmail? Oh, I see. So, okay, so they have all—they have my 3 million emails. Why don't they have a vector store where I can just like basic RAG, right? They're indexing the entire internet, uh, in real time. Like, I, I don't think my email is that big a deal, but yeah. My standard thing on memory is, it sounds like you are using, uh, Zero. There's also MgGPT, Neta, uh, which gave a workshop at my conference there, Zep, which uses a graph database, just open source, kind of interesting. And LangM from LangGraph, which I would highlight also—like, it's really interesting this developing philosophy that's—that people seem to be agreeing on—on a hierarchy of memories from semantic memory to episodic memory to—I think just overall sort of background processing—like we have independently reinvented that AI should sleep, Y, to, uh, to do that deep REM, uh, processing of memories. It's kind of interesting.

Yep. Yeah, that is—it's the other—I mean, just on the notion of memory and hierarchies, um, so, you know, I talked about, uh, the memory we're working on right now is at the, at the user level, and it's cross-agent, right? Um, but the other kind of one step up would be—so once again going back to this kind of hybrid digital teams—is that, uh, you can imagine to say, oh, well, my team, uh, has this kind of shared—I don't want shared with the world or anything other—this set of agents across this group of people—I want to have shared state, like we would have in a Slack channel or something like that—that should sort of exist as an option, right? Um, and the platforms, uh, should provide that, and, uh, the B folks—I should also mention—have mentioned that they're, they're working on that as well. So imagine being able to share, you know, selective conversations with people—like that's nice. Limitless has, I guess, voice-based shielding. I don't think they—is—I'm an investor in that too, by the way. So, full—

Okay. Trying to think about all the things I've said—investment—OpenAI, Perplexity, LangChain, Cohere AI, Limitless—a bunch of them. So, if I've said anything, by the way, I have no insider knowledge; I have no—I'm not trying to plug or pitch or anything like that. I think it's understood—where we're often like, you know, if you have skin in the game, you've probably invested, or you know, may—I'm not an investor in B, but I'm just a friend, and, uh, I think you should be able to speak freely of your opinions regardless.

Okay, we have some, um, miscellaneous questions that may be zooming out from agent AI. Sure. First of all, you mentioned this, and I have to ask: you have, you know, so many AI projects, you'll never get to. Yep. Uh, what's one or two that you want other people to work on? Oh, wow. Drop some—from your—other people to work on because you'll never get to it. Yeah. Yeah. You—what I need to do—I've had this thought before—so I have this is like maybe like pick one a week or something like that and give the domain away, uh, like have people submit their one or something like that. It's like if you can convince me that you have at least enough of an idea, enough, uh, like willingness to kind of commit to actually doing something, it's the ones that you keep mentioning but you, you, you, you haven't got into it for whatever reason.

Yep. Yep. Um, trying to think—like some of them I don't have the underlying business model. We're going to have—we're going to have to come back to this, maybe do a follow-up episode. I don't, uh—you're just not jumping to mine. You don't need the business model; just just scout a, uh—I think that's an interesting, by the way—pretty much all of them—there was an idea at the time; it's like it was one of those late nights—I like, "Ah, I could do this"—is a domain available—and now go grab it. Um, I'm trying to think what else I have on the AI space. I have a lot of like nonprofit, uh, domain names as well, for like, uh, nonprofit—like Open Graph, um—I'm not sure why things are not jumping to my head, uh, I, I, I have agent.com, which obviously is tied to agent.o. That's going to be big; that's going to be big. Oh my God, that's going to be like a 30, $50 million—it's going to be big. Um, it's—yeah, it's going to be—I think end up being bigger than, uh, chat.com, which—it has to—was 15.

Yeah. Yeah. Yeah. It's more work-oriented. Yep. That's interesting. Yeah. Do you want to talk about the chat.com thing? I would love just the backstories, like did you just call up Sam one day and be like, "I got the domain?" Did they kind of get back to you knowing that you had—

I'll give you—it's a, it's a good story. Back, uh, in the original ChatGPT days, uh, the first thought I had in my head—which lots of people had in their head—is that OpenAI is going to build a platform, and ChatGPT is actually just a demo app to show off the thing, and there's been precedence for tech companies that have had, uh, you know, uh, demo apps to kind of help normies understand the underlying technology, and even after the kind of the boost or whatever. So my original thought was, well, someone should actually create like an actual real product, and so I'm like—and that product should be called chat.com, because GPT is not a consumer-friendly thing at all; like that's an acronym—not—doesn't roll off the tongue. And so like, I'll build ChatGPT because I was just a demo app back then. So I, you know, got chat.com, and then as it turns out, ChatGPT is like a real product, and I was at an event here in San Francisco that Sam spoke at where he launched, uh, plugins, I think it was the, the announcement that that time Y, and that's the thing is like I had sort of suspected—it's like, okay, things sort of be like—there's no way that OpenAI is going to launch plugins for ChatGPT if they were not thinking of it as an actual platform. So it's not just about the GPT APIs; this is like a real thing. I'm like, "Crap! Like this violates the first rule of Dāreś, which is don't compete with Sam." Um, I knew when I bought the domain that there was competition for the domain, uh, there were, um, other companies looking to buy it. I, I don't know who they were; I had suspicions. Um, so I bought it, and then I'm like, okay, well, I'll reach out to Sam. It's like, "Hey, Sam, uh, I happen to have got—I don't know if you know—he was or wasn't, uh, kind of in the running or trying to acquire it or not—but I have chat.com. I'm not looking to make a profit; if you want it, you'll obviously do something much better, bigger with it. I don't want to be in the comp with Sam game," effectively is is what I said, uh, and so they did want it, um, and yeah, we struck a deal. Looks like it's been a very good deal if, uh, the valuations are, you know, to be, to be real. Yeah. Who knows? Who knows? It's one of those weird things like—

Yeah. The agent. domain evaluator said that laden.space is for between, between 5 and 15K. Okay. So does that feel right? Is—well, it's missed the—it's missing—so this is V1 of it; this one does not incorporate the transactional data. I have not published that one yet, uh, that's—and because it's also operationally very intensive, uh, that other one. But anyway, uh, we, we actually had it donated by a listener. Okay. So I don't know what the real cost is, but, uh, it's missing that—it's linked to an influencer, by the way. I also—Crew—Crew.ai, which I've offered—I'm an investor in—yes, I bought that, uh, and I've told him that like whenever you're ready, you let me know; I'll sell at cost. Uh, yeah. So yeah, I mean, that, that is some value-add. Since you may buy a lot of domains, what, what are your favorite domain buying tips apart from—have a really good domain broker, which I assume you have?

No, I actually don't. Uh, I do, I do my own deals. Um, God. Um, I have a like a very cards-face-up approach to life, um, so there's—so, you know, some people would tell you it's like, oh, well, if someone—they know that it's you're behind the transaction—you know the price is going to go up—sure—but it's still like willing seller, buyer, whatever—doesn't mean I'm going to have to necessarily pay that price, uh, it's like, okay, but the upside to it—because I always, you know, reach out as myself when there's a domain out there, um, and they can look you up, they can look me up, but then I also come off as like legit—like, okay, well, there's very few people are not going to return my email when I say I'm interested in a domain that they may have for sale, um, or had not considered selling, but you know, "Would you consider selling?" Uh, so yeah, and some of the like—so I own some of my favorites—I still own prompt.com, by the way; that, that could be a big one—and but I owned—and this is one, uh, I don't regret it—I went into a good—I owned playground.com, and the original idea behind playground.com was, at the time, OpenAI had their, uh, playground where you can play around with the models and things like that, right? It's like, okay, well, there should be a platform-neutral thing; there should be a playground across all the LLMs, then you can—and there are obviously products and, uh, startups that, that do that now—and so that was my original thing; it's like, oh, there should be playground.com, and you can go test out all the models and play around with them just like you can with, with OpenAI's GPT stuff. And then, uh, so was out there with, with playground—the company—been on the pod—yeah—and I think he reached out—might have reached out to me over, over Twitter or something like that—so we knew of, of each other. I've never—I've still never met him—and, and he asked me whether I would, you know, consider—and that was a tough one because I'm like, I actually have the business idea already in my head; I think it's a great domain name, uh, and it's like a really simple English word that has like relevance in a whole new context now, but once again, I took, took equity, so it's like, "Look on the bright side," that's like—I—so domains to get me into deals that I would never been able to likely get into other ways. So yeah. Yeah, we should securitize your GoDaddy account and just make it a fund—it's a fund—it's basically a fund. Yeah. Um, and by the way, so back to the kind of, uh, things—whatever—invested, uh, I don't know if it's public yet, um, but in a company that's going to treat domains as a fractionalized, uh, tradeable asset, because that's the kind of the original NFT in a way, right? It's like, okay, well, you—and then if you can make both fractionalizing, but also just a transfer—like right now it's so painful when you buy a domain, you go through an escrow service, and there's just all this—it's like I just want like instantaneous—like charge me and Bitcoin or credit card whatever it is, and then I should show up, and I should be able to route the DNS—like that should be minutes, not weeks or days. Um, anyway, so yeah, that's what ENS on Ethereum is basically—the same—they should bring it for normies.

Yeah. Exactly. They should bring it. Yeah. The—I can—and all of that is a, is its own, its own thing. I have a question on, on just the, you know, you keep bringing out the—your Sam Altman rule—one of my favorite, favorite, favorite—my first millions of all time was actually without you there, but talking about you. Okay, cuz, um, Sean was describing you as a fierce nerd, which I'm sure you, you, you were there, um, and I think Sam also is a fierce nerd, and, and he is—I was, I was listening to this Jessica Livingston podcast Y where—while she had him on—and described him as a formidable person. I think you're also very formidable, and I just wonder what makes you formidable, what makes you a fierce nerd, what, what keeps you this driven?

Yeah, Sam's fiercer and nerdier, just for the record. Um, but I think part of it is just like the strength of my conviction, I guess, like I'm, I'm willing to like work harder and grind it out, uh, more than people that are smarter than me, and I'm only slightly stupider than people that are willing to work harder than me, right? Like, I'm just the right mix of, uh, that kind of grind-it-out kind of work at it, stick to it for extended periods of time. If I think I'm right, I will latch out—latch on—and not let go until I can either like prove to myself that it's not—so even like the natural language thing, it's like, yeah, took 20 years, but eventually I got to a point where, uh, the world caught up, and, and it became possible. But yeah, I think—and part of it is, uh, I think this is partly—I think what makes me—like I'm a nice guy—sometimes they're the most dangerous kind, right? It's like, okay, well, I, I, I don't make enemies or whatever, but so my advice would be—my—this is my take on competition—I don't think of it as like war; I think of it as, uh, their opponents, all right? And this—not—it's like, it's, it's a game, right? And you can—and use whatever analogy—I happen to play a fair amount of chess; I'm a student of the game. That's partly, I think, what, uh, makes me effective. I'm solving for the long term, uh, so I'm kind of hard to deter. So for those of you out there looking to kind of compete with HubSpot, uh, good luck. No—I'm going to be here—been 18 years—I'm going to be here for another 18 years. So—but not that you shouldn't do it—it's a big market; I'm not trying to dissuade anyone—but yeah, I think like something I struggled with with is conviction. Yeah, you said you pursue things to conviction, but like you start out not knowing anything, yeah, and so how do you develop your conviction when there's—you, you find it along the way, or you, you stumble along the way, then you lose conviction, and then you stop working on it, you know, like how do you keep going?

The way I've sort of approached it is that, um, so I don't generally tend to have conviction around a solution or a product; I have conviction around a problem, uh, that says, "This is an actual real problem that needs to be solved," and I may have an idea for how to be solved, uh, you know, right now, and that I may be get dissuaded as like, "Ah, I'm not smart enough; technology is not good enough"—whatever—the constraints are—but it's the problem—I have conviction around it; it's like, oh, that problem still hasn't gone away, uh, so like I sort of filed away in the back of my brain, and I will revisit—like, okay, well, you know, the kind of the board changes, uh, and it changes really fast now with AI—like things that weren't—

Things that were impossible before are now possible. So you kind of go back to your roster of things that you believe or believed and say, maybe now, uh, now is the time; maybe then wasn't the time. Uh, but I'm a big believer in kind of attaching yourself passionately, with conviction, to problems that matter. Um, that and there are some that are just too high-flutin for me that I'm not going to ever be able to kind of take on. I have the humility to recognize that. Yeah, I feel like I need a, um, updated Founders version of a Serenity Prayer: like, give me the confidence to like do what I think I, I'm capable of, but like not to overestimate myself, you know. Yeah, uh, you know, anyway, uh, when you say board changes, how do you keep up on AI?

A lot of YouTube, as it turns out. Yeah, a lot. Um, okay, Fire Ship. I don't know what Fire Ship is. It's a current meme right now. Whenever Opening EY drops something, you know, they love this like live streams of, of stuff from on the Opening EY channel; the top comment is always, "I will wait for the Fire Ship video," okay, because Fire Ship just summarizes their thing in five minutes. No, I—so my kind of M.O.—so, by the way, I keep very weird hours. Uh, so my average go-to bedtime, uh, is roughly 2 a.m. Oh, boy. But I do get average 7, 7 1/2 hours in. Uh, I don't use alarm clocks because I, I don't have meetings, uh, in the morning, all, um, or try not to, at least. Uh, so my late-night thing is, uh, is I'll watch probably like a couple of hours of YouTube videos, often in the background while I'm coding.

Um, that's how you've seen our talks.

I have. Yeah, I've seen. Yeah. Okay. Yep. Um, and so there's so much good material out there, and, and the thing I love about kind of YouTube—this, by the way, in terms of like use cases and things agents that should exist that don't yet—I would love to—technology exists now to build this—is to be able to take a YouTube video of like a talk, say on late and space, or not, but on the, um, AI engineer event and say, "Just pull the slides out for me," uh, because I want to put it into a deck for use or whatever, or some form of, uh, kind of distillation or translation into a different—oh, I see—format. Pull the slides out of a video. Um, so I think that's interesting. I have—yeah, so, by the way, on the kind of agent thing, like one of the commonly used, uh, action primitives that we have is the ability to kind of get a transcript from a video, and that seems like such a trivial thing or whatever, but it's like—like if you don't know how to do it programmatically or whatever, if you're just a normie, it's like, okay, well, I know it's there, but I can copy and paste it, but like how do I actually like get to the, the transcript for you? And then, uh, getting to the transcript and then being able to encode it and say, I can actually, uh, give you timestamps if you have a use case that says, "Oh, I want to know exactly when this," because I want to create an aggregate video clip. This was the actual original, um, agent that I built for my wife that she wanted to pull multiple clips together without using video editing software, 'cause she wanted to have this aggregate thing, uh, she's on the nonprofit side, to like send to a friend. Uh, anyway, there are video understanding models that have come out from Meta, but the easiest one by far is going to be Gemini; they just launched YouTube support. Yep. So, um, they're doing good work over there.

In terms of like the coolest thing AI-wise recently, I'll say last week to 10 days, has been the new, um, image model, Gemini Flash Experimental, whatever they call it, uh, because it lets you effectively do editing, um, it, and just—and so my son is doing an eighth-grade research project on AI image generation, right? So he's kind of gone deep on, uh, Stable Diffusion and the algorithms and things like that. I don't know much about it, but one thing I do know—I know enough about Stable Diffusion to know why editing is like near impossible—that you can't recreate because it's like you can't go back that way; it's going to be a different thing because it's sort of spinning the roulette wheel another time the next time you tried to, you know, a similar prompt. And so the fact that they were able to pull it off—it's still, it's still a very much a V1 because, you know, if you—I know one of the test cases like, "Oh, take the Pot logo and replace the O, which like this kind of sprocket with a donut," and it will do it, but it won't size it to the degree that will actually fit into the actual thing. It's like, okay, um, but yeah, but that's where—so do you know the backstory behind that one?

No. Uh, Musta Mustafa, who was part of—so they had image generation in Llama 3, okay. Uh, lawyers didn't prove it; Mustafa quit Meta, joined Gemini, and didn't shift it, uh, and it is rumored—and that's all I can say—is that they got rid of diffusion; they did autoregressive generation, and I think it's been interesting, these two worlds colliding because diffusion was really about the images, and autoregressive was really about languages, and people were kind of seeing like how are they going to emerge? And on the Midjourney side, David Holtz was very much betting on text diffusion being, uh, being their path forward, uh, but it seems like the autoregressive paradigm is one—like, next token is—so Runway and Playground are doing like exceptional work on that. Uh, I don't know if it's autoregressive, but around kind of image editing and not just the kind of text-to-image and actually building like a UI for like a Photoshop kind of thing for actual generation of images versus—it's fasc—I thought diffusion was kind of dead, like there wasn't that much—it was just like bigger models, you know, higher detail, and now autoregressive come along, and now like the whole field is open. Yeah. Um, and I think like if there was any real threat to like Photoshop or Canva, it's this thing.

Just to wrap up the conversation, you have a great post called "Sorry, Must Pass," which, if I did the math right, you first wrote in 2007, the first version, right? And then you re-updated it post-Co—you mentioned you made a lot of changes to your schedule in your life based on the pandemic. How do you make decisions today, you know, in the—has anything changed like since you—because you updated this in 2022, and I think now we're kind of like, you know, five years removed from Covid and all of that. I'm curious if you made any changes.

Yeah, so that, so that post, "Sorry, Must Pass," was the issue that happened, um, is my schedule just—and life just got overwhelmed, right? It's like just—I just, uh, too many kind of dots and connections, and, and I love interacting with, uh, new people online. I love ideas; I love startups; there's a lot. But it turns out, uh, every time you say yes to anything, uh, you are by definition saying no to something else, um, just, uh, you know, despite my best, you know, attempts to change the laws of the universe, uh, I have not been able to do that. So that post was a reaction to that because what would happen for me, uh, would be when I did say no, I would feel this guilt because it's like, okay, well, whatever—it happened to me—it's like, "Oh, can you spend 15 minutes and just review this startup idea?" Or whatever. It's like, uh, and sometimes it would like be someone that was second-degree removed, like intro through a friend or something like that, and I felt, uh, you know, real guilt. And so this was a very kind of honest, vulnerable, "Here's what's going on in my life." So, so this is not a judgment on you at all, whatever your project or whatever your thing you're working on, but I have sort of come to this realization that I just can't do it. So I'm sorry, but I—so my default thing right now—and lots of people will disagree with this kind of default position—is that I have to pass because unless—and Derek Sivers said this really well—it's like either a "hell yes" or it's a no, right? So, and I'm—there's going to be a limited number of the, the "hell yeses," um, that I'm going to be able to kind of inject into my life. Um, so yeah, that, and that's—of all the blog posts I've ever written, that has been the most useful for me. So I, um, and so, and I send it—and I still send it out personally, right? I don't have—I don't automate my email responses at all yet. Um, don't do automated social media posts. Um, but yeah, that one's been very—and I—so I encourage everyone, wherever your line happens to be, I think this—lots of people have this guilt issue, and that's one of the most unproductive emotions, uh, in, in human psychology—is like no good comes from guilt, not really, unless you're like a sociopath or something like that. Um, maybe you need—anyway, you, you don't need more guilt.

Yeah, I would also say so, um, I would just encourage people to blog more because a lot of times people want like to pick your brain, and then they ask you the same five questions that everyone else has asked. So if you blogged it, then you can just—here—so one of things I'm, I'm working on—and there are startups that are working on this as well, uh, but I started before then—is like a D.A.R.E. that's just captures—and it's interesting—so that's one of the agents, um, on, on agent, uh, on the underlying platform. Oh, there, there's D.A.R.E.; it's out there; it's D.A.R.E. It's pure text space; no video, no audio right now, um, but, uh, the thing that's like I found it useful in terms of just the—how do I give it knowledge? So I have a kind of a private email address because a lot of the interaction that I will have—or if I do answer questions—because I—the other thing, by the way, I don't do any phone calls, like at all, even like no Zoom at all—get on Zooms with teams, but no one-on-one meetings, no one-on-one, uh, it just doesn't scale. So I've moved as much as possible to an async world. It's like I will—as long as I can control the schedule—like I will take 20 minutes and write a thoughtful response, but I reserve the right, uh, anonymously, with no attribution, kind of to share that, uh, either with my model or with the world, um, through a blog post or something, but it's been like useful because, uh, now that I have that kind of email backlog, I can go back and say, okay, I'm, I'm trying to answer this question, go to the vector store, um, and it's shockingly good, and, and I'm still irritated that Gmail doesn't do that out of the box. It's like they're Google. Um, I think it's, it's got to be coming now; it's—they—I think they're finally—the giant has been woken up. I think they, they kind of has gotten faster now, you know. It's one of the biggest giants in the world ever. Yeah. So yeah, when I first told Alesio, you know, you were one of her dream guests, I never—I never expected actually expected to book you because of "Sorry Must Pass." Yeah, so we were just like, "I'll send an email and like he'll say no and we'll move on with our day." Uh, so I just have to say like we're very honored that you just spend some time.

Thrilled to be here. Huge fan of—First Time, first-time guest.

Yeah, thank you for all that you do for the, for the community. I, I, I speak for a lot of them—you guys taught me a lot of, uh, what I think I know. So yeah, appreciate—

Yeah, I mean, uh, I am explicitly inspired by, by in, um, by HubSpot. Oh, thank you. Inbound marketing, uh, I think is struggle genius, and like the AI engineering is explicitly modeled after that. So like you created your own industry, you know, subsection of an industry that became a huge thing because you got the trend right. Yep. And that's what AI engineering is supposed to be if we get it right. Um, how do we screw this up? How do we square what up? How, how do I screw this up? How do we screw AI engineering up? Oh, um, you know, yeah, the—so the common failure modes, right, is, um, so the original thing that makes inbound marketing work, the kind of kernel of the idea was to kind of, uh, to solve for the customer, solve for the audience, solve for the other side, uh, because the thing that was, you know, broken about marketing was marketing was a very self-centered—"I have this budget; I'm going to blast you and interrupt your life and interrupt your day because I want you to buy this thing from me," right? And inbound marketing was the exact opposite—is like use whatever limited budget you have and put something useful in the world that your target customer, uh, whoever happens to be, will find valuable. Um, anyway, so the, the common failure mode is that you lose that—I don't think you will, but it's very, very common, right? It's like, ah, like now I'm just going to like turn the crank and squeeze just a little bit more like it's, uh, but you—you the right reason, I think, uh, folks like me, uh, you know, appreciate that community so much is you do have that genuine want to act, and there's nothing wrong with making money, there's nothing wrong with having an S.P., none of that, but at the, at the core of it, it's like we want to lift the overall level of awareness for this group of people and create value and create goodness in the world. Um, I think if you hold on to that over the fullness of time, uh, the market becomes more efficient and rewards that, that generosity. That's my kind of fundamental life belief. So I think you guys are doing—

Try for—

Thank you for your help and support.

Yeah, my pleasure. Yeah, and just a wrap in very DMes fashion, you have a URL for the "Sorry Must Pass" blog, which is sorrymustpass.org. So yeah, I thought that was a good nugget. Um, yeah, thanks so much for coming on.

Oh, thanks. Thanks for having me. [Music]