Topic

rag

Retrieval-augmented generation patterns that survive real users. Articles in this section cover chunking strategies, embedding model choice, vector databases (pgvector, Qdrant, Pinecone, Weaviate), reranking, hybrid search, query rewriting, evals, and the operational reality of running RAG in production. The focus is on patterns that work past the demo, including how to measure retrieval quality, when to skip RAG entirely, and how to combine retrieval with tool use in an agentic pipeline.