AI and security lab
An ideas lab on the frontier.
We think wild, build fast, and run experiments until they break or ship.
Method
Every idea runs the same loop.
Most ideas never leave the notebook. Ours get built, shipped, and broken on purpose.
-
1
Think
We test raw ideas with the people who would use them. Most don't survive. We build the ones that do.
-
2
Build
We make the smallest working version that proves the idea.
-
3
Run
We ship it, watch how people use it, and fix what breaks.
↺ then go again
Fields
What we work on.
AI
We build agents, the harnesses that run them, and the evals that show whether they work.
- Agents
- Harnesses
- Evals
Security
We attack our own systems before anyone else does.
- Attack paths
- Hardening
- Breaking things on purpose