RAG (Retrieval Augmented Generation)

LLM retrieves relevant documents from a knowledge base before generating its reply.

RAG ensures the LLM answers from 'your knowledge' instead of 'its training data'. Pipeline: (1) user question is embedded, (2) relevant documents retrieved from vector DB (pgvector), (3) LLM generates reply with these documents in context. LeadsUp's knowledge base is built on this pattern.