Field notes

Notes from the work of building durable AI systems.

Technical arguments, operating lessons, and research from the team building a sovereign runtime for human-like AI.

Start with the canonical work.

We publish when there is a complete argument or a system worth inspecting. These are the current references.

A practical case for using the best intelligence available without making critical systems dependent on any single provider.

How intelligence, hearing, voice, presence, action, and record become a governed end-to-end system.

Patterns for operating in Visca Cloud, private infrastructure, institutional networks, and air-gapped boundaries.