Retrieval-Augmented Generation (RAG)

Short answer

Retrieval-augmented generation (RAG) is a technique that grounds an AI model's answers in documents or data retrieved at the time of a query. The system first searches a knowledge source, such as product documentation or a database, then passes relevant passages to the model. This improves accuracy, reduces hallucinations, and keeps answers current.