---
title: "Retrieval-augmented generation (RAG)"
description: "A system that combines a model's learned memory with a retrievable store of documents at answer time."
canonical: https://aigovernanceengineer.com/glossary/retrieval-augmented-generation-rag
author: "Jorge García Aibar"
license: "CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/)"
doi: https://doi.org/10.5281/zenodo.22956197
version: "0.5.0"
updated: 2026-09-24
---

# Retrieval-augmented generation (RAG)

A system that combines a model's learned memory with a retrievable store of documents at answer time [1]. The corpus becomes behaviour, so it is governed like a model: versioned, carded, tied to the eval that tested it, with an entitlement check on what each user may retrieve.

- Developed in: [ch. 11, RAG systems](https://aigovernanceengineer.com/bok/ai-defined#rag-systems)
- Chapters: [ch. 11, AI Defined](https://aigovernanceengineer.com/bok/ai-defined) · [ch. 16, Fairness & XAI](https://aigovernanceengineer.com/bok/fairness-and-explainability)
- In the glossary chapter: https://aigovernanceengineer.com/bok/glossary#t-retrieval-augmented-generation-rag

## Sources

[1] Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks (parametric and non-parametric memory; arXiv 2005.11401). Lewis et al.. 2020-05-22. https://arxiv.org/abs/2005.11401 (verified: primary)