run-llama/llama_index

LlamaIndex is an open-source Python framework for building applications that combine large language models (LLMs), AI systems like GPT or Claude, with your own data sources. The core problem it solves is that general-purpose AI models do not know the contents of your private documents, databases, or company files. LlamaIndex provides the tools to load that data, index it (organize it in a way that can be searched quickly), and retrieve the right pieces whenever the AI needs to answer a question. The framework works through a concept called RAG, or Retrieval-Augmented Generation. When a user asks a question, the system searches the indexed documents for relevant passages using semantic search (matching by meaning, not just keywords), then passes those passages to the LLM so it can generate an accurate, grounded answer. LlamaIndex supports over 300 integration packages covering different LLMs, embedding models, and vector stores (databases specialized for semantic search), letting developers mix and match the components they need. It also supports building AI agents, programs that can autonomously take actions, call tools, and complete multi-step tasks. A companion cloud platform called LlamaParse handles document parsing and extraction for over 130 file formats, structured data extraction, and deploying production-ready document agents, though that is a separate paid product. You would use LlamaIndex when building a chatbot that answers questions about internal company documents, a search engine over a large knowledge base, an AI assistant for a software product that needs to reference documentation, or an automated research agent that reads and synthesizes information from many sources. The tech stack is Python, with a modular architecture where the core library handles the orchestration logic and separate integration packages connect to specific providers like OpenAI, Anthropic, Pinecone, or Chroma.