Technique
Retrieval-Augmented Generation
End-to-end pattern combining retrieval with generation to ground answers in fresh data.
Definition
RAG is the production pattern: embed user query, retrieve top-k chunks from a vector DB, stuff them into a prompt with the user question, generate a grounded answer with citations. The first step that most LLM applications take after a naive chat prototype.
Common use cases
- Knowledge bases
- Internal search
- Documentation chat