LlamaIndex Documentation
by LlamaIndex
The reference for building agents and RAG pipelines over your own documents.
Overview
LlamaIndex's documentation is the primary reference for building LLM applications that read your own data, and it now lives at developers.llamaindex.ai; the older docs.llamaindex.ai URLs 301-redirect there. The site is organised as Getting Started (concepts, installation, starter tutorials), Learn (building applications, agents, RAG pipelines, data extraction and evaluation), Use Cases (question answering, chatbots, agents, fine-tuning, multi-modal), Component Guides (loading, indexing, storing, querying, chat engines and agent deployment), an open-source community section, and an integrations catalogue. Four core abstractions carry the whole framework and the docs teach them in order: indexing, which structures data for LLM consumption; querying, through query and chat engines; agents, which are LLM-powered assistants given tools; and workflows, described as multi-step processes that combine one or more agents, data connectors and other tools. The integration surface is the reason most teams pick it, with 50+ LLM providers, 40+ embedding providers, 100+ vector stores and over 300 integration packages available through LlamaHub, and the docs pair the open-source framework with the managed LlamaCloud services, LlamaParse for vision-model document parsing and LlamaExtract for structured extraction. Installation is either a single pip install llama-index starter bundle or a slimmer core plus only the integration packages you need, which matters for production image size. The underlying repository has passed 51,000 GitHub stars and is actively developed.
At a Glance
- Topic
- RAG
- Level
- Intermediate
- Format
- Documentation
- Cost
- Free
- Duration
- Reference docs; the starter tutorial runs in ~1 hour, the component guides take 10+ hours
- Provider
- LlamaIndex
- Hands-on
- Yes — code/exercises
- Certificate
- None
What You’ll Learn
- ✓Load documents from files, APIs and databases using LlamaHub connectors
- ✓Build a working RAG pipeline from ingestion through retrieval to response
- ✓Choose an index type and vector store for your data shape
- ✓Construct agents that call tools and reason over multiple steps
- ✓Compose event-driven workflows that chain agents and data connectors together
- ✓Extract structured data from unstructured documents into typed schemas
- ✓Evaluate retrieval and response quality instead of eyeballing sample outputs
Highlights
- •The widest integration surface of any RAG framework: 100+ vector stores, 50+ LLM providers, 300+ LlamaHub packages
- •Workflows offer an event-driven alternative to graph-style orchestration, with agents as a component rather than the whole model
- •Component Guides document each swappable piece separately, so you can replace a retriever without rereading the tutorial
- •The starter bundle installs in one command while the slim core lets production deployments avoid pulling every integration
- •Documents the managed LlamaCloud path, LlamaParse and LlamaExtract, alongside the open-source framework rather than hiding it
Who It’s For
Best For
- ✓Engineers building retrieval-augmented question answering over private documents
- ✓Teams that need document parsing and extraction, not just chunk-and-embed
- ✓Developers comparing LlamaIndex against LangChain for a data-heavy application
- ✓Backend developers adding an agent layer over existing databases and APIs
Prerequisites
- •Intermediate Python, since the framework is class-heavy and async in places
- •A working idea of embeddings and vector similarity search
- •An API key for at least one LLM and one embedding provider
FAQ
What is LlamaIndex Documentation?
The official LlamaIndex documentation, covering the framework most teams use to put an LLM on top of private data. It walks from a starter RAG tutorial through indexing, querying, agents and event-driven workflows, then documents each swappable component and the 300-plus integration packages behind them. Use it when your problem is getting the right context into the model rather than training the model itself.
Is LlamaIndex Documentation free?
LlamaIndex Documentation is free to access.
What level is LlamaIndex Documentation for?
LlamaIndex Documentation is aimed at a intermediate audience. Recommended background: Intermediate Python, since the framework is class-heavy and async in places, A working idea of embeddings and vector similarity search, An API key for at least one LLM and one embedding provider.
How long does LlamaIndex Documentation take?
Expect roughly Reference docs; the starter tutorial runs in ~1 hour, the component guides take 10+ hours. Most learners work through it at their own pace.
What will I learn from LlamaIndex Documentation?
You'll learn: Load documents from files, APIs and databases using LlamaHub connectors; Build a working RAG pipeline from ingestion through retrieval to response; Choose an index type and vector store for your data shape; Construct agents that call tools and reason over multiple steps; Compose event-driven workflows that chain agents and data connectors together; Extract structured data from unstructured documents into typed schemas; Evaluate retrieval and response quality instead of eyeballing sample outputs.
Topics
Sources
This page was written from 3 sources, 2 on domains other than developers.llamaindex.ai.