Fig. 1: Standard RAG architecture — a user prompt is first processed by a Retrieval Model (e.g., vector search), fetching structured and unstructured data from internal sources; the relevant context ...