Signals

Structured inputs generated by AI systems that are evaluated by Governance for authorization decisions

Signals

Signals are the structured representation of AI-generated outputs used as inputs to Governance.

They are the only interface between AI systems and deterministic authorization.

---

System role

Signals act as the boundary layer between:

• probabilistic AI systems
• deterministic Governance systems

---

System model

AI Output → Signal Processing → Signals → Governance → Authorization Decision → Execution Runtime → Attestation
What signals represent

Signals represent:

intent
recommendations
observations
proposed actions

They do NOT represent decisions.

Signal structure

All signals must be:

schema-defined
normalized
validated
provenance-tagged

Unstructured outputs are not accepted by Governance.

Signal processing pipeline

Signals are produced through:

AI output generation
Signal transformation
Validation
Verification

Only verified signals are passed forward.

Why signals exist

Signals exist to:

isolate AI variability
standardize inputs for Governance
ensure deterministic evaluation
enable independent verification
What signals are NOT

Signals are NOT:

authorization decisions
execution commands
policy definitions
governance logic
Governance dependency

Governance only consumes Verified Signals.

Verified Signals → Governance → Authorization Decision

This ensures deterministic behavior.

Key invariant

AI can generate signals
Only Governance can decide
Only Execution Runtime can execute

Summary

Signals are the controlled interface between AI and Governance.

They ensure:

structured AI outputs
deterministic downstream evaluation
safe separation of intelligence and authority