#2297 Elad Gil backed 40 unicorns. This is next

19:36And eventually, you know, your coding agent will be able to deal with some of those dependencies, but is that the highest value thing for your coding agency to do? This is why internal tools often aren't, don't have giant teams at companies, right? It's because they want to deploy those engineers against product development, right? The bigger fish to fry. And then they realize, oh, we really need internal tools and we need internal tooling team. And sometimes those are amazing people and sometimes the people aren't as strong because they're really strong people who want to work on product, right?

20:03The coding agent thing may be similar in that you may only want to deploy them against the things that are highest ROI. You're going to have some limited budget for what you're doing. And the question is, is it rebuilding the landing page software or is it doing something else? And it's going to depend on the customer and the context and what it's worth and all the rest of it. And you're managing your resources to oversee those agents. All right.

20:22Elad Gil:So Ryan Carson, he goes, Andrew, I'm not using any standard ERP. My email software is something that I've created. It sends out custom messages at each phase of the drip campaign to the person. And I'm having, I forget what his chat bot's name is. I'm having Rosie, I think it is, write all these. And it's incredibly effective. In the future, everyone's going to do it. You're thinking that, you know what? Maybe in the future, everyone's going to do it, but they're not going to be like Ryan, who's like vibe coding this thing or somehow he's coding it. You're saying there will be a software provider that does it.

20:50Elad Gil:It'll be the new MailChimp. If you have that idea, don't give up on it and say it's going to be eaten away by Vibe coding software and people will create it themselves. Create that software. There's a future in that. Am I understanding you right? Yeah, I mean he may create the next MailChimp, right? I don't know. So that's the question is where does innovation come from and where does scale come from? And in general, what you find – I mean this happens all the time, right, where say you do have an internal tools team building something for you. This happened with a company that I helped out with quite early called BrainTrust, where, you know, they basically provide like an eval suite for enterprise amongst a bunch of other stuff, right?

21:27They have sort of a series of products now. But what they kind of started with was, can we help make eval really easy as you adopt the LLM so that you can change the model and you can see how do those changes propagate in terms of performance relative to some test stuff that you have? And, you know, a lot of companies are using them now in terms of sort of the leading companies that have adopted AI. And I think that that was a good example where I helped them with early customer calls and we'd call a customer. And those customers would often say things like, oh, yeah, I already have an internal team working on this.

22:01We don't need it. And then they call back three months later and they'd say, actually, we really need this. Our internal team, the resources should go to something else. This is important to us, but it's not our secret sauce. And not just that, you're now making up the number of 50 people and all of them are working on this one problem. and you're covering the whole surface area in ways we never will. And our three engineers should just be deployed somewhere else and we should just start using the software instead, right? And so I think there's a lot of examples like that back to, is it really worth your time to go and figure it out and all the nuances and all the use cases and all the lines of code and all the customization and all the performance and the cross-platform nature of it and cross-cloud nature and how you interrogate different data sources, you know, all that stuff.

22:42Like, why go do that yourself? So brain trust, I think, is a good example of this feature.

22:46Elad Gil:You're making me feel a lot more optimistic. I thought maybe you were knocking me down with this whole vision of toys not being possible. You have to have a rocket ship or else it's nothing. Now I'm seeing, okay, it is possible. There is a world where someone vibe codes something today, makes it more sophisticated tomorrow, adjusts it, and then they live and a customer keeps paying them monthly for it. You do still see that? Oh, yeah, 100%. Okay. What about this then? I think the big shift, though, that is a little bit under discussed is, and it's increasingly discussed now, is we're shifting from a world where you're buying monthly packages or you're buying seats for certain types of software and you're effectively buying units of labor.

23:28That's this agentic shift. And that, I think, is really interesting. And it also means that certain TAMs are much bigger than you think. So Decagon, where I'm an investor, is a good example of this, where they basically do customer support-related software or agents. And that's a good example where, you know, effectively, if you start paying for something on a metered basis or how many emails you reply or whatever it may be, you're effectively paying for units of labor. right you're converting something that used to be per seat sass like a zendesk and you're converting into how do i actually help customer support agents reply to things that scale much faster because the ai is effectively helping them do the responses right you're shifting labor sources in some sense over to more and more ai and that's going to happen throughout the whole services world and services economy and that to me is a really fundamental shift because that that's software eating into other parts of the economy that it didn't used to exist in specifically people uh well uh specifically certain types of highly repetitive uh language-centric jobs

24:35Elad Gil:okay you know i have one other thing that i'm that i'm going to circle back on what if the future of software for businesses is not anymore buying software like on a monthly per seat basis or per agent basis what if it's hiring someone like alex lieberman's company to create the software for you and it's no longer your team getting diverted but it doesn't cost them that much because maybe his company is doing this for three other businesses and so instead of buying not the ryan ryan carson is creating a divorce software because his sister went through tough divorce so he's got a divorce ai but if he were going to productize this email marketing customized solution maybe people wouldn't sign up to that maybe they would say i like that hey alex build the same thing for me.

25:22Elad Gil:And then Alex builds it for five businesses. He keeps adjusting and he keeps improving it based on what he's learning. And is that the future? Well, it sounds like it's just repeated, building repeatable software then. I don't know. What's the difference? Where's the line? Is it five customers versus a thousand? Is it, you know, why doesn't he just start selling that to everyone? So, you know, it just sounds like a different size customer. Now, if you look at certain types of industries, government, certain really big institutions, et cetera, they always did this they're still doing it right the um i don't know if it was the fbi or cia or somebody spent like 80 million dollars to build an ats they hired a third-party consulting firm i don't know if it was accenture deloitte someone built them an ats for like 80 million dollars they could have just bought greenhouse or in a more recent era ashby or something i see and so that's always existed that you have these big these people who are willing to pay but the ability to create software faster isn't didn't change all that and by the way i'm pushing back because i like your optimistic point of view you're basically saying there's a lot more opportunity here than i saw but i'm saying it's both i'm saying it's both i'm not i'm not saying it's a monolithic world i think there's a world where people will buy certain things that are repeatable have big surface area etc and then there's going to be all sorts of stuff that people make custom bespoke vibe coded etc or whatever you know the long-term i mean the long-term substantiation of vibe coding is really good code built by agents alone, right?

26:45With some human oversight, but much less than exists today. And so at some point that isn't even vibe coding, right? Because vibe coding almost has this negative connotation of it being kind of a bit more loosey-goosey, right? And so absolutely all that's going to come and all that's going to happen. That doesn't mean software goes away, but it does mean that there's going to be way more custom stuff. There's going to be way more creators. There's going to be way more people building stuff. I think that's great that's good for the world okay I asked you a couple of days ago I said hey

27:17Elad Gil:do you want to just share your screen and show me something cool that you're working on you go I don't know I don't have time for that I didn't think you'd have that response I thought you'd go Andrew I cannot wait for you to see this thing that I'm working on do you have something that if you had a little more time you would you just would be ripping out of your skin to show me what are you what are you doing that's like that I mean there's a bunch of stuff that I'm involved with that I think are really cool, interesting things. And one thing that I might even be working on is working with OpenAI and Anthropic and Eleven and a few other folks on taking the world's most important books and using machine translation to translate them into languages that represent over 80 % of humanity, creating audio versions of that in every language that's supported and basically just providing it as a free resource on the internet.

28:10And so we're calling that Alexandria after the great library of Alexandria. And so that'd be an example of something that I've been working on where we're using all the tooling and all the rest of it. And then there's kind of more day-to-day stuff like using cloud code or code work for different things or using cognition. We started using cognition in different ways quite early on. We've been doing really interesting things with open AI

28:30Elad Gil:and deep research and you know so like what what are you personally doing that are you doing any of this on your laptop yeah like what what are you personally doing that this is so exciting that you can't wait to bring me like if i was sitting in your office you go yeah i'll come see this well i mean a lot of what i've been doing is like scraping and interpreting data in different ways for different areas um i started doing this i mean i guess i started doing this a long time ago uh but there's a couple things that i'll show you maybe in a month or two when i'll be a little bit more baked, but I've been doing a lot of, um, kind of go scrape this information and let's start interrogating it.

29:04And that's been really interesting. And, you know, one example of something that I looked into, um, is, uh, ADHD data and diagnoses and, um, what's shifted because we basically went from a world where, uh, ADHD was something like one in, you know, thousands and thousands of people to now it's about 3 % of the population in kids gets diagnosed with ADHD. And so the question is, what caused that shift? And you can start knocking out factors that are claimed as the culprits. Parental age actually isn't that much of a factor. Like a bunch of stuff that's claimed to be the driver of all this stuff isn't.

29:39And the punchline is really it's a shift in incentives. And then it's a shift in incentives will be things like, hey, if you have ADHD, you get an extra two hours on the test or whatever in school. And as a teacher, you get rewarded for, hey, you're helping neurodiverse populations or whatever. um but it also includes um uh a shift in how diagnosis is defined a shift in terms of who can provide the diagnosis when the state of new jersey is something like 60 percent of adhd diagnoses in kids actually have never uh the kids have never taken any form of test it's just somebody deciding that they have it like a teacher just saying oh i think you have adhd and then they're classified that way uh so there's a lot of these sort of societal level drivers that are happening that are unrelated to you know the actual spectrum there um and so when you're doing this you're doing this yourself what are you doing what are you doing it yourself in

30:30Elad Gil:like if i were to look at your laptop what would i see you messing around with on your free time to do that it's a mixture of uh clod opening eye deep research gemini deep research it really depends on what model is available at the time and what sort of tooling i need and do i need certain things graphed or interrogated in different ways or do you're asking go get me this data tell me and then you're starting to ask questions when you say interrogating it's just you telling it to get the data and you're asking it to to analyze the data in some cases clean it in certain ways it's a bunch of stuff normalize these tables relative to each other i mean all the stuff that you'd normally would be a little bit more painful to do so i do love that you're doing the alexandria project i really feel like what's happening in tech is that there's we've taken our eye off of the helping humanity ball and we've kind of used it as an excuse for everything we're doing trust me whatever i'm doing is going to save the world and you'd given this this uh interview where you said past i forget what it was business leaders however you describe them they used to build monuments they used to actually like support the local arts you'd walk around and you would be somehow enveloped in a thing that they created and yes you might have known them as the entrepreneur who built the trains but they're also the people who are creating the park that you're sitting in and sure we have not done that in in tech and i like that you're doing the monument i sent you an article to show you how you should be pissing people off and feel comfortable because even the pyramid at the Louvre did that.

31:51Elad Gil:And I think that that's the way to go. But I especially think you've got this ability in tech beyond monuments that is even more important. Well, yeah, I think maybe what happened is there's an era where people thought the way to have societal impact, through capital at least, is through foundations. And so all these various foundations got set up and one could argue a subset of them have done great work and then a subset of them have potentially been detrimental to society in all sorts of ways in terms of the programs they've pushed or promoted or you know kind of called for um and what got lost as part of that where so much money and energy went into societal change and social engineering and you know trying to change the way our society works instead of saying how do we think about public beauty or how do we think about art or how to you know and so it's just um it's almost like we had a memetic shift in where capital got allocated relative to society-related projects.

32:45And I would argue some of it was incredibly well-intentioned capital that had the opposite effect of what it wanted to. You could see that in education or a few other areas where the capital wasn't always constructive and potentially was quite destructive. And then that almost drained, I think, effort away from the arts. And one could argue that could be the mindset of the people involved. You know, tech people really value education, while maybe people who are in finance in New York are exposed to and have art as something that they focus on more. And so there's also been a shift of, well, it's like a shift in societal outcomes and the background interests of those cultures relative to each other are different, right?

33:25Tech people are different from finance people in all sorts of ways. Okay.

33:30Elad Gil:How would, if you were looking at all these amazing tools that are created right now to create other tools and businesses. And you were starting out today. I know you hate these types. I don't know that you hate these types of questions. I hate these types of questions. And I especially hate to impose it on you because you're somebody who's such a big thinker, right? Rocket ships. How do you get people to live longer? But at the same time, I'm curious about how you see the world. And I want to see that understanding by seeing this little example. If you had all this, what would you start? What would you build?

34:00Elad Gil:How would you think about the fresh new idea to get into yeah it's it's hard to answer that because there's a lot that i think should be built right now and so and and there's a wide range of stuff i mean it's everything from new foundation models for materials like what periodic is doing for physics for variety of areas so i think there's a whole like set of models that nobody's built yet or that i should say people are building but it's very early that i think could be incredibly impactful and interesting and just like cool you know like it'd be super interesting to reinterpret physics through the lens of like a you know a model or a modern sort of architecture model um so i think stuff like that's super interesting i think if you go up a layer or two there's so many things to do on the application side again i think there's a real dearth of consumer products and there's a ton to build there and all sorts of interesting ways actually um me and somebody on my team ran this program at Stanford probably about two years ago, maybe a little bit more now, three years ago, where I think GPT-4 had just come out and we said, we'll give you free compute.

35:06We don't want anything in exchange. We just want you to run experiments and build cool consumer stuff. And we had about a dozen teams from Stanford go through it. It was all undergrads and grad students and CS and AI. And we'd have like a weekly check-in. And so that was a big investment of time, which was like, we'll check in, we'll help you with what you're building, we'll give you free compute and free model access, and let's just see what you come up with. And so I just think there's so much stuff to do. It's such an exciting era, you know.

35:35Elad Gil:You're also incubating. How do you find the ideas that you want to incubate? Is it in, this would be cool in the world, or here's a problem that I see and I need to solve it? It really depends. You know, I think it's a big range. I think often it's, hey, we've come across this problem sometimes by accident, and should we do something against it? You know, we've considered, and I say we, because like I have a group of people I work with now, considered, should we go and buy a company and then transform it with AI? But I mean, at scale, right? Is there something really big that you should work with, because they already have all the customers, they have the baseline platform, and really, they haven't layered in AI and can you grow the market dramatically or the use cases dramatically by adding certain things so I think there's everything from like incubate and get things started and up and running and all the rest which was Brainco and BrainTrust and a few other companies over time on through to do you buy a really late stage thing that's I mean it's a public company and you try and shift some of the focus of that company so So there's just so much.

36:45And then there's foundation models, right, that we talked about.

36:48Elad Gil:What's an example of a publicly traded company that you could buy and improve with AI that you decided not to? I don't want to say anything that we've specifically looked at. How about one that you're not, just to give me a sense of how you look at the world. Let's say one could look at what? Warner Brothers? No, that's pretty big. New York Times, publicly traded company, can't really buy them because of the way their stock's set up. Okay. Yeah. You know, a good example, and one could argue Nirvana has been doing this on the travel side, is you could imagine buying certain types of travel agencies or other folks and then revamping a lot of what they do, everything.

37:27And again, Nirvana has done some of this customer success and support. They've publicly talked about revamping with AI, you know, how you actually deliver services to people changes, how you think about trip landing changes, like a lot of stuff shifts if you view it through an AI lens. Interesting.

37:42Elad Gil:Okay. All right. I see where you're going with this. All right. Let me close out with this. We've been talking a lot about today. What's tomorrow? Like, I loved how you found AI as the place to invest in because you'd read the papers, you'd met some people. I especially love how you decided that you're going to invest in. And then once Google got out of the military, right, they were out of servicing the military. You said somebody in here is doing something smart and doesn't have a place to do it. I'm going to go talk to them. What are you seeing today that's exciting for the future? Yeah. You know, the honest truth is when you look at these technology waves, they are not obvious until they're obvious.

38:22And then everybody starts looking for the non-obvious thing again. But the obvious thing is what you should do. okay and so i think for a good example that would be everybody proclaimed that social products were over like four times right first there was the wave of like um the really early social wave was like myspace and friendster and multiply and all these companies and then facebook became the really big thing and then twitter happened two years later i guess linkedin happened a couple years before and then right around after twitter happened i always said okay social is largely saturated and then suddenly you had Instagram and you had Pinterest and you had Snap and you had people like it's over and then you had TikTok and you had you know it just keeps going and some and then now I think it's kind of more done right until there's some AI centric thing or whatever but it took a while it took like a decade right SAS took a decade to really see itself through fintech took a decade I think AI is going to take a decade and for the next decade plus there's going to be really really interesting things happening and you know there's other areas that I think are interesting i think energy is interesting i think you know defense is still interesting you know i think there's a variety of areas that are still interesting but ai will continue to be interesting for a long time so and i get it you're saying look andrew stop looking beyond ai we're

39:35Elad Gil:not even a decade into this thing it's super exciting the early days it's the early early days well i mean i'm also doing i continue to do uh some things in space and defense and you know different forms of American resiliency and all this stuff. But, you know, I've been looking at energy for a while now, you know, years and years of looking at it, not doing much, but everyone's mother something interesting. But I think fun, I've done crypto before, like years and years ago, although I haven't done anything recently. So I think there's just these waves and some of the waves are really interesting for a long period of time.

40:09Some of them kind of collapse after some, some moment and the big companies come out of it and you're done. AI is one of those that's going to go for a long time. because basically everything software is AI.

40:19Elad Gil:Tell me, can you reveal one thing that you're working on that you're so excited about that you shouldn't talk about, but you can't even help talking about it? Maybe it doesn't have to be the most biggest secret. Come on, this is Silicon Valley. Nothing's a secret. I mean, honestly, I do want to get a sense of what you're working on that you're, I like to see your passion. What are you passionate about that you're working on in AI now? Yeah, it's a mix of, again, looking at some of these other types of foundation models besides language. Like what? it's um you know there's there's bio there's chemistry there's material there's there's like all these different categories right okay um so i think something like that are really interesting um there's looking at uh you know things at the application layer we talked about consumer we talked about um you know roll-ups we talked about services transformation we talked about vertical applications like it's all that stuff like it doesn't change it's the same stuff what do you see consumer that's exciting it doesn't seem to me like you i don't know i felt like maybe the beginning that you weren't as excited about and i don't see you in consumer tech much what do you see in consumer that's exciting yeah um i used to do consumer tech i was you know i invested in airbnb and instacard and my stuff quite early and then i worked at twitter for a while i sold a company to them um but then i think there was a gap where there wasn't that much and um you know one could i think a lot of the kind of prosumery things are a mix of consumer and professional use cases.

41:43So early mid journey was both of those things, right? You do it for fun and for art, but you also use it professionally. Perplexity is that way. Open AI is that way. You know, I think, I think there's a lot of blurring. You know, my hope is that we see more and more companies tackling big problems. I mean, you know, there's stuff around email, there's stuff around productivity, there's stuff around personal agents, there's, you know, there's a lot to do. Now the question is, does a big company do it or does a startup do it? And that's the hard thing with consumer right now, because the difference between now and 15 years ago is there's a handful of companies with such large distribution that they're the natural homes to do new things.

42:24Because even if they're five years late, they can just cross sell it. And that's often what happens. In the early days of a startup, you're competing with other startups. In the later days of a startup, you're competing with incumbents. And so the question is, when do incumbents wake up? And is there a gap that's big enough that between the incumbent waking up and you're working on something, you can actually get it done to enough scale that you survive?

42:45Elad Gil:But maybe it's so small that they just don't care about it. And then eventually it's too late for them to care about it. Yeah. And the too small may still be quite large, right? If you think about these companies, if you're a$1 to$4 trillion market cap company, you need a line of business that's, I don't know,$5,$10,$20 billion in revenue eventually to move that number at all for you. And that means that something that looks like it's a couple hundred million dollars isn't that interesting. And so you're going to miss stuff that looks like a couple hundred million dollars is really$20 billion in revenue, right?

43:19And that's the kind of loophole you need to find as a founder. If you're building something that an incumbent should be doing.

43:27Elad Gil:I had an entrepreneurship professor who told us to think like Mao. He said Mao didn't just march on the capital. He went to little hillsides, the countrysides. He went and he built up his people there. He preached. He built up his forces. And then once he was big enough, he could take over the whole country. And what most people do is say, here's the biggest thing. Let me just go after that because then I'll be king. Yeah, I usually need some weird wedge in. like uh vinod casa has a good saying that your market entry strategy is different from your market disruption strategy and i think that's very true like what's the wedge in and sometimes it's head-on and sometimes it's more adjacent to the side and most of the time it's probably more adjacent all right i like this conversation i feel like we've only scratched the surface of friendliness and happy which i kind of like because i've listened to a lot on your podcast and other people's podcasts you're very serious and very big and very intimidating person and i like the happy go lucky part of you.

44:20Elad Gil:The, the, the optimist part that I get to see. Um, all right, let, let me close out with a fun thing. What do you do for fun? What are you doing? Yeah. Yeah. It's a ride of things. Um, you know, love hiking, working out, uh, yoga, um, travel, you know, it's kind of standard stuff. All right. Thanks so much for doing this interview. I feel like we need to, did you try? I appreciate it. I was, I was going to come up with something like really fascinating and fun for you to do but i don't i don't think you're a kite flying type of person not a kite flying what is it called a kite surfing type of person no i'm not it's too much it's too much gear yeah too much overhead you have to show up a skier are you i snowboard i snowboard all right snowboard yeah not very well but snowboard yeah thanks brother congratulations i'm looking forward to talking with you again hopefully in person with a whiskey yeah sounds good thanks