What is Retrieval-Augmented Generation (RAG)?

Retrieval-Augmented Generation (RAG): Definition

Retrieval-Augmented Generation (RAG)

Also: RAG, retrieval augmented generation

Retrieval-augmented generation (RAG) is the technique behind most AI search: instead of answering only from memory, the model retrieves relevant documents at query time and grounds its generated answer in them, then cites the sources it used. It is why fresh, reachable, extractable pages can be quoted right away.

Updated August 11, 2026

RAG separates what the model knows from what it can look up. The retrieval step is where your content enters the answer, which is why being in the index and easy to extract matters more than waiting for a model to be retrained on you.

The name describes the sequence: retrieve relevant documents, augment the prompt with them, then generate the answer. It was introduced in a 2020 paper, Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks, by Lewis et al. at Facebook AI Research with University College London and New York University. Engines retrieve a candidate pool, evaluate it for authority and agreement, then generate an answer citing the strongest sources.

In practice the retrieval step is rarely one query. Google documents a "query fan-out" technique in AI Overviews and AI Mode, issuing multiple related searches across subtopics before synthesizing — so a page can be pulled in by a subquery the user never typed.