CocoIndex — инкрементальная индексация данных для RAG

★ 11.4K

CocoIndex is an incremental data-indexing framework that turns codebases, meeting notes, inboxes, Slack, PDFs, and videos into live, continuously fresh context for AI agents and LLM apps. Reach for it when your RAG system or agent needs not a one-off dump into a vector database but a pipeline that picks up changes and reindexes only the delta, so context never goes stale and nothing is missed. You describe the pipeline declaratively in Python in minutes; it processes data in parallel, scales, and writes embeddings and metadata into your database for downstream semantic search. Its focus is building and maintaining a fresh context index for AI (ETL for RAG), not being the vector database itself (embeddings go into Qdrant/Milvus/pgvector and the like) or a ready chat-over-docs app: it is the data-preparation layer that search and agent apps rely on. In a RAG stack it sits in front of the vector store.