framework
LlamaIndex
LlamaIndex is a framework specializing in data connectors, indexing, retrieval, and agentic workflows over private and enterprise data.
- Python
- TypeScript
- framework
Website: docs.llamaindex.ai/
GitHub: github.com/run-llama/llama_index
Languages: Python, TypeScript
License: MIT
Best for
- RAG-heavy and data-centric agents
- Enterprise document workflows
- Indexing and retrieval pipelines with agent layers
Pros and cons
Pros
- Excellent data ingestion and index abstractions
- Strong RAG + agent combination stories
- Broad connectors
Cons / trade-offs
- Can be more than you need for non-RAG agents
- Architecture choices still require discipline
Overview
LlamaIndex excels when agents must reason over your data: ingest, index, retrieve, then act. It is often chosen for knowledge-worker and enterprise document agents.
Fit signals
Pick LlamaIndex when retrieval quality and data plumbing are the hard part of the product.
Related reading
Agent Memory Explained
Working memory, thread memory, long-term memory, and RAG—how memory works in agentic systems and how to govern it.
How to Build a Research Agent
Step-by-step design for a web research agent: goals, tools, planning, citations, evaluation, and safety limits.
Tool Use Pattern
How agents call tools reliably: schemas, selection, error handling, permissions, and verification of side effects.