Key Concepts

Understanding these terms will help you navigate GOVERN and configure it correctly for your organization.

Assessment

An assessment is the result of evaluating one inference against your active policies. It contains a score (0–100), any violations found, and the action taken (pass, flag, or block). Every inference that flows through GOVERN produces an assessment.

Policy

A policy is a collection of scorers, thresholds, and enforcement rules. Policies are attached to AI systems or system groups. When an assessment runs, GOVERN applies the policy attached to that system and returns a verdict.

Violation

A violation is raised when a specific scorer exceeds its configured threshold. Multiple violations can exist on a single assessment. The enforcement action (log, flag, block) is determined by the most severe violation’s enforcement mode.

Energy

Energy is GOVERN’s measure of governance health over time. It aggregates assessment scores, violation rates, and remediation velocity into a single signal. High energy means your governance posture is strong. Dropping energy is an early warning of governance debt accumulating.

Drift

Drift is detected when a model’s behavior deviates from its established baseline. GOVERN monitors distribution shifts in output patterns and alerts when drift exceeds your configured threshold.

Monitoring Agent

A monitoring agent is a lightweight worker deployed in your infrastructure that continuously scans for AI systems. Agents report discovery events back to GOVERN and feed the inventory.