Backed by Y Combinator!

We got into YC. The first week is behind us.

A few weeks ago, an email arrived that we had read in our heads many times before it existed. DeepMark is part of Y Combinator's Fall 2026 batch.

We started in Podgorica with one idea, that if AI is going to speak, its voice should carry proof of who is speaking. We grew up as a company in Belgrade, where our research team still sits. And now, for three months, the two of us live in San Francisco.

The first week was a lot. New place, new city, a room full of people who are all building something they cannot stop thinking about. The batch opened with Sam Altman on stage.

What he said stays in the room, but we walked out with the feeling that the bar is exactly where it should be. The questions here are simple, and they do not let you hide. What did you ship this week? Who did you talk to? What did they say? That is the point, and it is why we wanted to be here.

Nothing about what we build has changed. The watermark goes inside the voice, it is inaudible, it survives phone calls, and the person on the other end can check who is really speaking. What has changed is the pace, and the number of people who now want to see it work on their own audio. We are happy to help!

Thank you to everyone who believed in a small team from Montenegro and Serbia long before anyone in Silicon Valley knew our name. You know who you are. We are here because of you.

If you build voice agents and want to see the mark on your own audio, write to us.

Slavko and Kosta