Analyst Briefing

Agentic Governance, defined.

The analyst-grade definition of the category, in the precise vocabulary the rest of the briefing uses. Agentic Governance is governance FOR agentic systems — any system where authority to act is delegated across humans, AI agents, and the tools they invoke.

The five primitives

  • Act — something an agentic system does that has consequences.
  • Authority — the delegated permission that licensed the act.
  • Purpose — the declared reason the authority was admitted.
  • Scope — the bounds the authority was admitted within.
  • Evidence — the signed chain that resolves the whole to finality.

The category boundary

In scope: who held authority for an act, within what purpose and scope, and whether that is provable. Out of scope (and consumed as input): whether a model is fair, statistically valid, or free of prompt injection. The boundary is what keeps the category coherent — and what makes it distinct from the adjacent categories on the landscape map.

In closing

The definition compresses to one essence: every act has authority; every authority has finality.