Retrieval-Augmented Generation

RAG combines a vector database with an LLM: relevant document chunks are retrieved and fed into the prompt.

Pipeline

Ingest → embed → store → retrieve → generate.

Related articles