← Back to Glossary
Retriever
Phase 03 · RAGcore-concept

The component in a RAG system that searches a database and fetches relevant document chunks for a query.

Why it exists

LLMs have static knowledge cutoffs and limited prompt windows. The retriever fetches targeted facts from your private knowledge base.

How it works

Converts user query into vectors or keyword queries, searches an index, and returns top matching document chunks.

Where you'll use this