Enterprise AI Agent Governance: A Layered Approach (Build, Deployment and Runtime)

### Executive Summary This document presents a three-layer governance model for enterprise AI agents—build-time, deployment-time, and runtime—detailing typical controls, configuration risks, and runtime enforcement mechanisms (e.g., prompt injection detection, data leakage prevention, tool-call validation). It emphasizes that secure agent scaling requires all three governance layers working together to prevent misconfiguration, mitigate dynamic runtime threats, and ensure accountability across the agent lifecycle.