EMO-64
A reference architecture for systems that observe, detect change, evaluate permitted transitions, decide, act, verify outcomes and record evidence.
Overview
EMO-64 is a reference architecture for governed autonomous systems. It treats the model as one component inside a larger operating loop rather than as the entire product.
The architecture defines seven stages — observe, detect change, check permissions, decide, act, verify and record evidence — and makes each stage an explicit, inspectable boundary. An agent does not act because it produced a plausible sentence; it acts because a permitted transition was evaluated, an action was selected within policy, and the outcome was verified against intent.
The work is research, not a shipping product. The goal is to make the boundaries around an AI system as engineered as the model itself: permissions are typed, decisions are logged, evidence is durable, and every action can be traced back to the observation that motivated it.
Related Projects