LangChain
Open-source framework for Python and JavaScript that lets you build AI agents and LLM applications from models, tools and data sources.
What it does
LangChain is a library for Python and JavaScript/TypeScript for building AI agents. With create_agent you combine a language model, a list of tools and a system prompt into an agent. A unified interface lets you switch between providers such as OpenAI, Anthropic, Google or local models via Ollama without rewriting much code. According to the vendor there are more than 1,000 integrations for models, databases and tools. Agents run on LangGraph, can persist intermediate state and ask a human for approval before taking actions. The paid LangSmith platform adds tracing, evaluation and hosting.
Who it suits
Development teams that want to build their own AI features into software – for example a chatbot with access to company documents or an agent that queries internal systems. Without programming skills you won’t get far with it.
Data protection for businesses
The framework runs in your own environment; your data goes to the model provider you connect, and that provider’s terms apply. If you use LangSmith, you choose the US or EU region (Google Cloud in Europe) when creating your account. According to its terms of service, LangChain does not use customer data for training. A data processing agreement is part of the terms of service and can be signed in a pre-signed version. Self-hosting and hybrid deployment are available in the Enterprise plan.
Limits
Costs for the model APIs always come on top. The free LangSmith plan is limited to one person and 5,000 traces a month. For simple single calls to a model, a framework adds extra complexity.