Glossary · Industry term
Retrieval-augmented generation
Also known as: RAG, retrieval augmented generation
An LLM application pattern that grounds generation in retrieved documents fetched at query time. Originally introduced by Lewis et al. (Facebook AI) 2020. Now the default architecture for enterprise LLM deployments that need verifiable sources.
How this publication uses it
RAG is necessary but not sufficient for enterprise reliability. The retriever can return wrong context; the LLM can ignore the context it receives. RAG-poisoning — adversarial documents seeded into the corpus — is the indirect-prompt-injection variant most common against enterprise knowledge bases.
Articles that analyse this term
Primary sources
- Lewis, Perez, Piktus et al.. Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks· 22 May 2020