langchain

Why we included this project

LangChain is the framework most Python teams reach for first when they start building LLM applications, and for good reason: it handles the plumbing of wiring chat models, tools, retrievers, and data sources together, so you don't have to hand-roll every integration. It gives you a standard interface for chat models, embeddings, and vector stores, plus a large collection of provider and tool connectors, which means swapping an OpenAI call for a local model or adding a new data source rarely requires rewriting your application logic. On top of that it layers higher-level agent patterns and a durable runtime for stateful, long-running workflows, so the same code can carry you from a quick prototype to something you can actually operate. If you want fine-grained control over agent orchestration, you can pair it with LangGraph; if you just need a fast, proven agent template, the built-in create_agent path gets you there. When you're deciding how to structure a new LLM product or standardize an existing one, this is the framework most of the ecosystem assumes you already know.