The substrate, not the wrapper.

Most AI products are thin shells around someone else's cloud model. I build the other thing: deterministic Rust systems with local models doing the reasoning — monitoring that runs unattended for months, agents with real internal state, knowledge engines built on actual physics.

I'm Bludlock — I build deterministic Rust substrates with locally-hosted LLMs as a bounded reasoning layer: NATS-coordinated daemons, structured-output inference, no cloud dependency. Everything here runs on hardware I own. This is where I show it and write about how it's made.