LlamaIndex — RAG и агенты над документами

★ 37k

LlamaIndex is an open-source framework in Python and TypeScript for building AI applications over your own data, primarily RAG and agents that answer from documents. Reach for it when you need to connect your data to an LLM: load and split documents, compute embeddings, put them into a vector database, and build search and question-answering with source citations, plus agents that traverse that data. It provides connectors to hundreds of sources (LlamaHub), convenient abstractions for indexes, retrievers, and query engines, without tying you to a specific model or store. Its focus is a specialized 'data → RAG/document agent' framework for developers, not a visual no-code tool or a vector database (embeddings go into Qdrant/Milvus/pgvector): it centers on connecting and querying your own data, whereas LangChain is a more general LLM-app builder.

• #RAG