AI Agent Governance
The infrastructure and processes for governing individual AI agents — including identity, trust levels, access scope, constraint enforcement, and activity monitoring.
AI agent governance manages individual AI agents as entities within an organisation's governance structure. Just as human employees have defined roles, authority boundaries, and accountability, AI agents need:
- Identity: A named, trackable entity (not an anonymous process) - Role: What the agent is responsible for and what it's expected to do - Trust level: How much autonomy the agent has (shadow, preview, active, autonomous) - Access scope: What systems, data, and tools the agent can access - Constraints: What the agent is explicitly prohibited from doing - Activity monitoring: What the agent has done and whether it's operating within boundaries - Accountability: Who is responsible for the agent's actions
AI agent governance is distinct from model governance (which governs AI models in training and development). Agent governance applies to deployed agents that are taking actions in production.
How Constellation handles this
Constellation provides a complete AI agent governance framework: agent roster (identity and role), trust levels (progressive autonomy), governance gate (constraint enforcement), and traces (activity monitoring).