RAG evaluation measures whether a system retrieved the right passages and whether the answer stayed faithful to them. It covers relevance, citation quality, and regression when documents or models change.
Codoop ships an evaluation harness with golden questions before an assistant reaches users, then reruns it when content or prompts change.
Related: /ai-services/enterprise-rag