Agent frameworks

Agent frameworks are the software stacks that orchestrate Large Language Models (LLMs), external tools (APIs, code execution), and memory to create autonomous, goal-driven AI applications.

These frameworks provide the modular components necessary to transition LLM prototypes to reliable, production-grade systems. Key players include LangChain, AutoGen, and LlamaIndex: each specializes in a critical area. LangChain offers the general-purpose toolkit for chaining complex reasoning steps and tool use. AutoGen excels at multi-agent orchestration, enabling a team of specialized agents (like a Coder and a Tester) to collaborate on a task. LlamaIndex focuses on Retrieval-Augmented Generation (RAG), efficiently connecting LLMs to private data sources like PDFs and SQL databases for grounded, factual responses. They are essential for building agents that can plan, execute, and iterate on multi-step tasks using external data and actions.

https://docs.langchain.com/