Rohan Kodialam on Sphinx AI and building copilots for enterprise workflows

any watch on the planet. Seriously, any watch. Our next guest is in the reream waiting room. We will bring him in. and to the TBPN Ultradome and continue today's show. We have Rohan from Sphinx AI, I believe. What's happening? Welcome to the show. Great to meet you. Thanks for having me. Thanks so much for hopping on.

Uh, give us a breakdown. Introduce yourself. What are you building? What's the news? Absolutely. Yeah. So, I'm Roan. I'm one of the co-founders and the CEO of Spinx. Uh, Sphinx is a company that is solving AI for data.

Uh, and specifically, we're razor focused on the idea that large language models in AI today are really good at text. They're good at code, which is just a type of text, good at images. What they're not good at is the type of tabular semistructured time series information that makes most of the world work.

Everything from supply chains to high finance to uh manufacturing. Uh we're bridging that gap and delivering co-pilots that can think about data natively and help people get from piles of raw information to valuable business ideas, what we call alphas in finance, but can be extended to almost any industry. Interesting.

Who's the first customer in finance then? Yeah, first customer in finance. Don't know if I have the rights to name them, but we do have I mean like the the architect, private equity, hedge fund, quant frequency. So yeah, so someone who's like analyzing a stock on a quarter by quarter basis almost. That's right.

Y that makes quarter by quarter. Well, faster than quarter by quarter really. Um we're working with a lot of quants. Um so people who are maybe day by day or even faster than that. Sure. But yeah, the hedge funds are the target market initially.

they are able to solve these problems where they find y a single insight that can make billions in a year. That's interesting. Yeah. And then I imagine that if you go any faster, the customer wants to write their own o camel if we're talking about Jane Street. And so it's for specific quant that's looking for Yeah.

dayby-day analysis. Interesting. Um so uh what is the what is the go to market like? I I imagine that uh maybe it's more of like an enterprise software deal. you're actually going outbound, meeting with folks, stake dinners, uh signing deals, or is this kind of like general sign up on the website? Both, right?

So, we are we are going to people that maybe not stake dinners yet. Don't know if we have that budget, but we certainly are um talking to a lot of different companies across industries, but it also is a bottoms up motion. Um there are a lot of people, myself included, in my past role. I used to be at Citadel.

I I ran a quant team. uh I was very frustrated at the lack of tools like this to help me get from idea to value. So we do have quite a robust set of uh data scientists, quantitative researchers who are just using the product off the shelf, but we're also making those enterprise sales as well. That's awesome.

What is uh give us your read on the talent wars between the Citadels of the world and the Frontier Labs and now Sphinx, of course, because I'm sure you're competing. Um yeah. Um no, we we've we've gotten people from hedge funds, not from Citadel.

I do have a non-solicit but we have gotten people from other hedge funds and we have gotten uh people um to take our offer over say anthropic or open AI. Uh look like it comes down to what you're optimizing for.

I think if you're going to these foundation labs you're betting on AGI and if you win good for you that's a great bet. Um hedge funds are betting on almost nothing. I think if the AI bubble pops tomorrow it'll make a hell of a lot of money and that's great for them.

Um and really for us we're focused on our specific application. So it comes down to what you're passionate about. Is it is it money? Is it AGI? or is it just like razor sharp focus on solving one problem and that that's who we hire. What's your take on uh on Microsoft Copilot in Excel?

We've talked to probably four different companies that are building Excel add-ons. Uh there's there's that world. Uh how does that fit into your strategy? How uh are you doing something different or or do you see that on the road map? How do you think about that?

Look, like I think Excel has been the tool for data for like what however long it's been out there. But we don't see that as being the transformative layer. Um the kind of serious data work, the kind of data work that makes people billions of dollars is done on a slightly different stack.

A stack that can scale a little bit more that translates more naturally to production and and we really want to target that. And you know, Excel is easy to use, but what's easier than Excel is writing prompts. Um so over time, I'm not saying I'm killing Excel today, right?

Over time, I see that as kind of becoming a less relevant layer, almost just the interpretability layer on top of something like Sphinx. Uh, actually getting work done in Excel, I feel, is going to become a thing of the past. We're going to be able to build much more serious models with a lot more financial impact.

It's exciting. How do you charge for Sphinx? Because it it feels like uh, you know, or the best companies create a lot more value than they capture. AUM, but uh, just 2% of AUM. Just give me 2% of your AUM Citadel. We're good. Yeah. Your entire I'd love that. hire management.

Uh no, but it's like, you know, if you help if you help somebody make if you help people make money, they're happy to pay you uh quite a lot for that. Yeah. Look, um we we don't only work in finance. We we're already quite horizontal and it's very hard for us to tell the exact nature of the value we're delivering.

Um but some of our customers of course are open with others like hedge funds, they're not going to even tell us like what they found.

Uh so the the kind of a way of like capturing that value you know it's it's hard it's hard for us but really for us the focus is on building something transformative and um you know this is step one as we're able to get more and more autonomous and we're able to eventually say you know we can solve large scale data problems in a couple seconds fully autonomously I think the value becomes enormously clear.

So we're not overly focused on that right now really right now we just charge for usage. Um we want to deliver people something that they use when they use it they have to pay us to get that service. That makes sense. Uh well, thank you so much for taking some time out of your day to come talk to us.

Yeah, it was great to meet you. Congratulations. We'll talk to you soon. Have a great rest of your day. Uh and I will tell you about Wander.