00:02:36 · Danny Fortson
[00:02:36] S2: Hello, and welcome to Danny in the Valley, your weekly dispatch from behind the scenes and inside the minds of the top people in tech. I'm your host, Danny Fortson. And this week, we are talking AI, but not, you know, like sexy AI things like digital girlfriends or robot lawyers, but a startup that is actually being used by companies to do complex work, which makes it different from most AI companies. So the company is called Hebia, and they make AI agents that can carry out a whole range of of tasks, like, say, helping asset managers understand the market and their exposure to it or saving hundreds of hours of human labor by, say, doing document review for law firms, for example. Anyhow, the company is four years old. They just raised a $130,000,000 in July, and they are led by George Civulka, who is all of 26 years old. He's a former academic. He quit, Stanford four years ago to start the company. And what is interesting is that so far, they seem to have built something that real grown up companies are actually paying for. And as discussed a lot on this pod, most of the AI out there today is kinda garbage or they're just toys or they're pretty one dimensional things that can do kind of one thing. And Hebe appears to have created a more capable general reasoning engine that you can throw a lot of different things at and critically get paid for doing so. So we talked to George about his path from his upbringing in New York as a kid of immigrant parents, his younger days as a mathlete, getting his first gig at NASA as a teenager, convincing Stanford to fund his research on how coral reefs reason. That is a true story. And it involved, as you might guess, lots of time on a boat off the coast of Palau because why not? We also talk about how he met Peter Thiel getting his first check and how AI is, in his view, is about to transform the economy as we know it and do so pretty quickly. So we cover all of that and much more. It's a really fun conversation. So I will now hand you over to my chat with George Savolka of Hebia. Enjoy. So I've been like pretty much everybody in this world. I've been covering AI kind of incessantly for the past couple years and here and there before that. And one of the things I think was interesting about you guys is that, well, obviously, you raised... Was a $130,000,000 recently, which is a lot of money.
00:05:21 · Danny Fortson, George Sivulka
But also, it seems like you guys are trying to do something that a lot of these other AI startups say they're doing but aren't. And so I really wanna kind of, drill into that because it feels like there's so much excitement. There's a lot of stuff that's been funded, and then a lot of companies being like, this is kind of cool, but I don't wanna pay for this because this doesn't work or doesn't work reliably. So can we start with what is the big idea for Hebia, and why have you called it Hebia? [00:05:56] S4: Those are two completely different questions and stories. [00:05:59] S2: They are.
00:06:00 · George Sivulka
[00:06:00] S4: And they're and they're they're fun in their own regard. So I'll I'll I'll handle one first and then we'll dive into the second because the second is actually, pretty interesting. But you're right. There's thousands of startups that kind of are building a chatbot or a search engine or training large language models. They all have one fundamentally very large problem. And that problem is that they're they're doing a single step or they're doing a single action. And the way that humans work, you... You're right to call it work, has multiple steps. It has a complex process. Yeah. And so if you if you think about all these other start ups, they're they're building these transactional one off relationships where Heavya and our big idea is humans are much more complex than a single step. The way we work is much more complex than a single step. And the interface to AI should go way beyond a chatbot or way beyond a search It should actually be able to handle that complexity, to handle multiple steps. And a good example would be, you know, even evaluating whether or not something is a good investment. Yeah. It's probably like the most important question in the workplaces of most of New York firms. And if you ask a chatbot that, it'll say, I can't answer that question or a search engine will find you what the CEO of the company thinks. But the way that you actually evaluate what a good investment is is you'll decompose it into how strong is the management team, What is their revenue and revenue growth relative to peers? And a thousand other questions from considerations about their customer concentration to long term impacts the industry. And Heavier will actually do that complex work. Instead of just trying to find what the CEO is saying or searching for something that exists, it can do work that doesn't yet exist. And this is a bit of the advent of agentic AI or, like, AI that can do work.
00:07:49 · Danny Fortson, George Sivulka
[00:07:49] S2: Right. So like a kind of a smart coworker. Exactly. [00:07:53] S4: If if current AI startups and apps are not actually driving real value because they're just doing a single step, that's the worst coworker of all time. [00:08:02] S2: Right. [00:08:02] S4: Even an intern, you give them an arbitrary task and the intern will go in and try to solve that task over and over and over again. They'll execute a bunch of different steps. They could be totally wrong, but you're not gonna have to stand over the intern shoulder and say, okay, now do this. Okay. Now do this. Yeah. Okay. Now do this. [00:08:18] S2: And so are you building your own large language model? Are you building this on top of OpenAI or Anthropic or one of these other ones? Like, what is the actual technology that you're building to actually create something that is a bit more agile and a bit smarter? [00:08:35] S4: So Hebia fundamentally believes that the value is not in the model layer. We think that it'll actually become relatively commoditized. [00:08:44] S2: Agreed.
00:08:44 · Danny Fortson, George Sivulka
[00:08:44] S2: Agreed. [00:08:45] S4: We actually used to train, and our bread and butter was training, and I'm I'm formerly an academic. And there was a moment in 2020 or 2021 where I I realized that because the value wasn't going to be accrued in the model layer, we should totally stop and work on some other secret sauce. And as I thought more and more about the way the industry would evolve, it seemed like, okay. There's this really important technology, but there wasn't a product that was there. And that product would need infrastructure. It would need orchestration. It would need a... An elaborate interface that would allow people to do that more complex work. And so we stopped training models, and that was like a... It tore my heart apart, you know, as a former academic. And we started actually building a product from scratch, which sits on top of any large language model. We're LLM agnostic, and our customers sometimes even use their own large language models in heavy. [00:09:37] S2: Right. [00:09:38] S4: And it orchestrates how those LLM should be called. A really good kind of similar example would be Microsoft Excel. You don't care what the back end powering calculator or computer or database is. You just care about the way that your calculations get orchestrated, the way you interface with them, that it's a container for that work. And similarly, our customers don't care about the l one behind it.
00:09:38 · George Sivulka, Danny Fortson
[00:09:38] S4: And it orchestrates how those LLM should be called. A really good kind of similar example would be Microsoft Excel. You don't care what the back end powering calculator or computer or database is. You just care about the way that your calculations get orchestrated, the way you interface with them, that it's a container for that work. And similarly, our customers don't care about the l one behind it. [00:10:01] S2: If it's, for example, Lama from Meta or Mhmm. OpenAI, you know, ChatGPT, whatever, and take... Choose any of them. They're all slightly different. They all have slightly different capabilities. They all have slight, you know, different weaknesses, different strengths, etcetera. Does that not directly impact what Hebia can do? If you're just saying, okay, we're just gonna put this on top of any LLM. Just give us an LLM. We'll make this work and make it smart and make it really capable as you're saying. But if you're doing it on top of a... One that is not good at one thing or better at another, does that not directly impact what you guys can do? [00:10:40] S4: I think the the question is, does the car drive if the engine is bad? [00:10:45] S2: Yeah.