The substrate layer for
autonomous computing.
The problem
Every current agentic system is an improvisation. A language model is given tools. It calls them in sequence. Someone writes orchestration code to handle errors. Someone else adds a logging layer. The result is a system that works until it doesn't — and when it fails, nobody can say exactly why.
What we are building
stratum is a principled architecture for the layer between an autonomous agent and the world it acts on. That architecture has five concerns: workspace surface, unified multimodal model, action surface, escalation primitive, and capability tokens.
The audit store
Append-only. Tamper-evident. Every action carries a provenance tag: which agent, which context, which reasoning trace, which human approval if any. State is a projection of an event log. You can always reconstruct how you got here.
Contributing
The architecture is not frozen. The implementation does not yet exist in any coherent form. What we want right now: people who read this and have a strong opinion about where it is wrong. Open an issue. Disagree publicly.