← Back to glossary

RAG and search

Retrieval-Augmented Generation (RAG)

A technique that retrieves relevant evidence from approved sources and supplies it to a generative model before it answers.

Business example

An employee asks about an expense policy and receives an answer grounded in the current policy document with citations.

How it works

1.The question is converted into a search representation.
2.Relevant permitted passages are retrieved.
3.The model answers from that context and cites the sources.

Common misconceptions

RAG means uploading PDFs to a chatbot.

Production RAG also needs ingestion quality, permissions, retrieval evaluation, citations, and freshness.

Related concepts

Embeddings Semantic search Vector database

Related reading

RAG is not uploading PDFs: permissions, freshness, traceability, and reliable sources

Sources

https://arxiv.org/abs/2005.11401

Reviewed: 2026-08-04

Reviewed by: Javier Chulvi Bernad · LLM Engineer · Madrid