Retrieval-augmented generation (RAG)

Combining a retriever over documents or tools with a generator LLM so answers can cite fresher or private context.

RAG retrieves relevant chunks from an index, database, or corpus and conditions LLM decoding on that context, reducing reliance on memorized parametric knowledge alone.

Trade-offs: retrieval quality, latency, prompt fit, and attribution; grounding still needs human review for high-stakes facts.