WeKnora
Document-to-RAG pipeline in a box, from a team that runs it at Tencent scale.
WeKnora takes raw documents through parsing, chunking, embedding and retrieval and hands you three surfaces on top: a question-answering RAG, a reasoning agent, and a wiki that updates itself as the corpus changes. It is Go-based, ships with a single-binary deployment story, and is unusually opinionated about the boring parts most frameworks leave to you — incremental reindexing and document versioning.
Best for: Teams that need a self-hosted knowledge base over internal documents, especially with Chinese-language corpora
Deploy: Self-hostable
RAG in Production Playbook
Chunking strategies, reranking setups, evaluation harnesses and the seven failure modes that kill every RAG demo in week three.