Breaking, Building and Betting on AI Panel | Sigsum 2025

Hey everyone, uh thanks for coming. It's great to see you. Uh this is a conference for product builders and so I'm very happy uh to be up here with three amazing product builders. Uh Karen Deepep who is the CEO of Character. Malta who is the CTO at Versel and Tuin who is the founder and CEO at base 10. Um thank you all for coming. Um maybe we can start and just uh I can ask each of you to just give a little bit of your background, a little bit of your story. How did you end up here? We'll go.

Happy to start. Uh currently pan Mike mentioned uh actually there's a mic version of the story in this. I used to be at Microsoft big companies and then finally uh I left to join meta Facebook back then and that was thanks to Mike and the first thing he advised was like change your address email address from KSR and Hotmail to Gmail address you can get a job in the valley. So I did big company Microsoft came to Facebook went even smaller was at Brex for three years

I was CPO uh president there and then the last six months been at character AI it's been an interesting journey from 100,000 to to 40,000 to 800 people now 50 so nice so I'm Malta originally from Germany been here like 12 13 years most of the time actually at Google um originally worked kind of in the infrastructure seller. Then at some point for some reason they let me run part of the search engine. And so I was like very deep in this like you know let's test every shade of blue type of

world. And so I I I I quit Google I joined Versell and I'm like how do you do experiments? Um and I was so was very happy when stats eventually kind of came to the market right after. It was good timing. Um so yeah now now I'm over Versell a CTO um kind of trying to make AI accessible to everyone. Yeah. Hi Tuan um CEO of base 10. Um I

think different to a lot of product builders. I started my career as a banker um in New York City. Uh got bored of that very quickly. Uh ended up joining a healthcare setup in Boston and kind of fell in love with early stage companies. So for the last 15 years I've been doing machine learning in a bunch of different forms at early stage companies. Uh eventually got the founder bug started the company started based in 2019 which were a bit early to ML and AI and you know the kind of indust industry

kind of showed up in front of us and you know for the last three years we've been hyperfocused on production grade inference. We work with a bunch of fast growing companies. Amazing. Um maybe in the reverse order we can just uh you you all touched on it a little bit but just uh maybe share a little bit about exactly what your companies do. So you guys are the leading AI inference platform. You work with folks like a bridge and open evidence and cursor and clay and a whole bunch of companies but maybe you can uh take a moment and just sort of describe

what you guys do. Yeah. Um yeah, so three years ago when open source started to take off and after the chat GBT moment, uh a lot of people started having pretty serious inference problems and we just happened to be building um just for a few years prior and so that's what we do. We do production grade inference for you know the fastest growing companies and um like a bridge and cursor and open evidence and clay and really what we focus on is speed, reliability and developer experience and how do we kind of let our customers focus on what is

unique about them and let us abstract out as much of the boring infrastructure as possible. Yeah, I mentioned already I've been with the cell for three and a half years and and the company I thought I joined um was making web frameworks and um and then it was making the infrastructure so that you can run them in production in a way that's super fast, super scalable and and very much like at an infrastructure level designed for personalization and and you know modern delivery