Retrieval-augmented generation: chunking, embedding, storing, top-k retrieval with metadata filters, adding results to the prompt

A large language model only knows what was in its training data, and it may confidently invent answers about your private documents. Retrieval-augmented generation (RAG) fixes this by finding relevant passages from your own data at question time and putting them in the prompt, so the model answers from supplied facts. RAG has an ingestion pipeline and a query pipeline, and the data services from this domain sit in the middle.