Hi, I'm Donovan, building AI action security for the physical world
Hi, I'm Donovan, founder of BNCR in Austin. We're building security for AI actions in the physical world, the layer that looks at what an AI system is about to do before it reaches real hardware or real-world processes. The work is early, so I'd rather keep this short than oversell it.
My stack right now is Claude and ChatGPT for thinking through design, Ollama for running local models, Claude Code and Codex for building, and Vercel for shipping. ElevenLabs is in the mix too.
What I'm still figuring out is which physical-action scenarios create the sharpest need for this kind of check, and who should be our first users. I'd love an introduction to one or two people who have deployed AI agents near machines or facilities, plus blunt feedback on how we describe the problem.
