Back to glossary

Term

LlamaIndex

Data framework for LLM applications, specialized in indexing, enriching and retrieving private data — at the core of many RAG architectures.

LlamaIndex — in more detail

LlamaIndex (formerly GPT Index) is an open-source framework for Python and TypeScript that focuses on connecting your own data to large language models. It bundles the typical RAG building blocks — data connectors (PDF, Notion, Confluence, SQL and many more), chunking, indexing into vector or graph stores, query engines, retrieval strategies — under a unified API.

Example / practical context

A classic LlamaIndex pipeline loads documents via SimpleDirectoryReader, builds a VectorStoreIndex, and lets a QueryEngine answer questions against the index. Composable indexes let you nest several sources — for instance a hierarchy of summaries over original documents. With LlamaParse and agent components, the framework also moves toward complex multi-step workflows.

Delineation from similar terms

LangChain is more general-purpose and ships a broader set of agent, tool and memory abstractions. LlamaIndex deliberately focuses on the data path: what to do when your knowledge source is bigger than the model’s context window. The two frameworks can be combined — LlamaIndex serves as the retriever, LangChain orchestrates answer generation.

Entdecke mehr