How does RAG work?

Question

How does RAG work?

Accepted Answer

RAG first retrieves relevant chunks from a vector database, then injects them into the prompt for the LLM to generate a grounded answer.