Definition

Measurement of AI quality and failure modes against the real task, such as accuracy, completeness, latency and review effort.

Why it matters

This concept requires clear ownership, integration and controls in a real implementation.

Related terms

AI GovernanceRetrieval-Augmented Generation (RAG)Model Runtime