00:15:07 · Elad Gil
[00:15:07] S0: It all gets absorbed. Yeah. Mean, they work on a railroad. Yeah. It's very durable. So I think I think there's a few different types of models. Right? So there's the the core language foundation model, like the LLMs, etcetera. That's mainly what we talk about. And I think there... Two, three years ago, everybody thought, okay. This is gonna go over time. And for things, you need some form of persistent memory. And you need to have these sort of actions, and you need people to tie into these different... This is to integrate with these 30 thin... You know, like, I think it was kinda clear what to do, and it's just a matter of time to go do it. Right? We knew reasoning was coming because we'd have to be seen reasoning in other contexts. And... You know? So I I think I think a lot of it has been reasonably predictable. And, like, I remember talking to Eric Steinberger who runs magic. And three years ago, he was talking about super long contest windows being incredibly important for code because then you could put the whole code base in, and they could work across it. And, you know, so, again, I think a lot of the the smartest people had all these ideas that announced for Tintia. And then similarly, I think there's a very clear roadmap in terms of what's to come. For other areas, I think there's lots of really cool stuff to do. So are... You know, what are the best models for physics simulation? What are the best models for materials? What are the... You know, you kinda go through a whole by circle or area by area, and I think there's super interesting stuff to do. Now that may be separable for economic value. Right? And that's my big question about, some... A subset of the biomodels. I think there's broader things people are doing. But, you know, if you look at the cost of developing a drug, for example, let's say 1,500,000,000 and it kicks fifteen years, tens of millions of that is preclinical work. That's the molecule. And the other 1,450,000,000 is a clinical trial. Right? Now you may be able to impact the success of a clinical trial based on the molecule. You look at Admed and other aspects of developing drug, etcetera. But fundamentally, I I... You know, those companies, in my opinion, many of them will end up just being drug development companies. And the question is, do they have real systemic advantages?
00:17:02 · Elad Gil, Aditya Agarwal
And so those feel like tougher categories and maybe physics or language or image gen or other areas. And so I think part of it is like, what are you doing? How advanced is it? How interesting is it? How how much utility does it happen? Is it overlay is like, what is the economic model that you're dealing with relative to that type of model? [00:17:19] S1: Maybe two questions rolled up into one here, Elad. So I mean, how... I I I think that over the last six weeks, it's probably been more encroachment from the foundation models into a number of, like, different domains. Thought going out for profit, thought for legal, thought for finance, and obviously of all the coding related things. So I guess the two questions here are like, how do you as a startup, if you could start something, how do you think about what is it in the effective blast radius of what these models can do? And then secondly, I guess related to that is what are the durable moats that an application or a product can have? And I think one of the things you said in the past, which actually resonated with me a lot, was that data is maybe overrated as a mode. Everybody talks about we have we have proprietary data, and we can, know, do better, like, context window stuffing. And that feels a little thin to me and an argument. So different questions folded up into the same thing there.
00:18:17 · Elad Gil
[00:18:17] S0: So if you look at the history of technology, every time you have a new platform, it forward integrates into the most valuable application in the platform. So for example, Microsoft OS board integrated what became Word and Excel and PowerPoint and Access, and they basically killer bot companies that were doing that. And then they port integrated in the browser, and that was a famous sort of fight with Netscape and then I've used Google in the two thousands port integrated into the biggest vertical search categories. Right? They had local, and they did travel, and they did finance. And so, this always happens. And so it's not odd that the foundation model companies or labs will forward integrate into the biggest applications starting with code. Code in particular, though, also has this interesting attribute, which if if you have a very good coding model, it helps you generate the next model faster. Right? And that's because it helps you write code for it. It helps you with data labeling. It helps you with a bunch of stuff. And so, you know, one of the hypotheses in terms of how we get true lift off into a singularity is you have models, Boolean models, Boolean models, and then you just take off. Right? And so it's not a surprise that these... The labs are focused on that. It seems like the the single queerest specter of speed of change. And eventually, we have, like, evolving models and evolutionary trees, and you spawn 50 versions of the same thing, and there's some utility function abrasive against and all sorts of crazy shit that, you know, seems likely to happen at some point.
00:19:36 · Aditya Agarwal, Elad Gil
[00:19:36] S1: So what is the world? It's great. We'll be all, you know, we'll be all at [00:19:42] S0: the edge of this. I'm sorry. I got a little bit distracted. [00:19:45] S1: I was gonna be like, here is Satya. Eat it bright. Do you have me happy?
00:19:45 · Aditya Agarwal, Elad Gil
[00:19:45] S1: I was gonna be like, here is Satya. Eat it bright. Do you have me happy? [00:19:50] S0: They should not ask me questions like this. This is a loaded question. I don't know whether they're gonna afford an interview. Yeah. Exactly. So I If you look at durability, there's strategic ways that you can be durable. And by the way, if you look at... You know, the whole wrapper around a model thing, don't really buy. Because if you look at a SaaS company, is this a wrapper around a SQL database? And I'll fix it. Those things just just bite. Right? I think usually what you need is you need to build a multiproduct company. So there's three theories. One is there a system of record? You're basically where core data about a thing is stored in all its attributes. And so, you know, that's a workday for all the people working at your company or whatever. And so then you keep all the dozen applications around it, and it's very, durable. That's that's one hypothesis in terms of what creates durability. I think the single biggest thing is just, can you build a dozen different products that you're cross selling roughly to the same customer or user? They're deeply integrated, and therefore you have multiple workflows. And I think then you're very defensible. So for example, if Harvey has two dozen different workflows for different legal applications and use cases, that's defensible. If they're just, you know, let's chat with the legal document, they're not very defensible.
00:20:57 · Aditya Agarwal, Elad Gil
[00:20:57] S1: It's a good observation. I actually haven't... Maybe I've said this publicly before, but that's interesting because it's a very, know, kind of what Parker always talks about with the tech startup. [00:21:06] S0: It's Rippling. It's HubSpot. It's actually Microsoft. [00:21:10] S1: It's a lot more doable today because you can just crack that stuff out. [00:21:13] S0: Yeah. It's what's the surface area of your product, and I think it's very under discussed and is often discussed as a revenue driver, which it is. And, you know, you have different wedges in that you can cross sell against. You can cross sell to the same account. It's hugely defensible. Right? Because usually what happens is, a founder will start a company that build one product, and they'll win because a product is 10 x better, and that'll get in distribution. And then they, keep going, and they forget the fact that now that they're the incumbent and they have all these customers that can build a product that's 80% as good as the best thing and still win because they're just cross selling. They've already done the security review. They've already done the purchasing. They were... You know, they've gone to procurement. They've gone through everything legal. And so it's really easy to start cross selling, in which case it becomes very defensible. Right? It's really hard to rip out a dozen things instead of just one thing. [00:22:00] S1: I was gonna say this, why we don't really like to sub Teams. [00:22:04] S0: Teams is such an interesting example of that. Right?