Naive retrieval — embed the query, grab the top-k nearest chunks, stuff them into the prompt — is how almost every RAG system starts, and how almost none of them stay once real users start asking real questions. This piece is a survey of the retrieval patterns that consistently show up in

Enterprise data rarely exists as isolated pieces of information. Contracts, policies, employees, vendors, regulations, and operational records are connected through complex relationships. Traditional Retrieval-Augmented Generation (RAG) systems typically split documents into chunks, convert them into embeddings, and retrieve the most semantically relevant chunks for a query. This approach works well