Skip to content
Deploris

RAG (Retrieval-Augmented Generation)

A pattern where an LLM answers using passages retrieved from your own documents at query time.

RAG stands for retrieval-augmented generation. A retriever finds the most relevant passages in your knowledge base, then an LLM composes an answer using only those passages and cites the source. RAG is easier to update than fine-tuning and safer to audit because the source of each fact is visible.