Glossary
Retrieval-augmented generation
A technique in which a model first fetches relevant documents and then uses them to write its answer.
Retrieval-augmented generation (RAG) combines a search step with a language model. When a question comes in, the system retrieves relevant passages from a set of documents, such as a company's files, a legal database, or the web, and gives them to the model to base its answer on.
RAG helps keep answers current and tied to specific sources without retraining the model, and lets the system show where information came from. Many enterprise and government AI tools work this way.
It does not guarantee accuracy. The system can retrieve the wrong document, or the model can misread or ignore what it retrieved.