What is LangChain?

LangChain is an open-source framework and ecosystem for building applications and agents powered by large language models (LLMs). It provides modular components and abstractions to connect LLMs with external data sources, tools, databases, and workflows, enabling the creation of chains, agents, and more complex applications in Python and JavaScript/TypeScript.

The ecosystem includes several key open-source frameworks: LangChain for rapid prototyping with pre-built agent architectures and integrations to any model provider or tool; LangGraph for building reliable, stateful, and controllable agents with low-level orchestration, persistence, streaming, and human-in-the-loop capabilities; and DeepAgents for highly autonomous, long-running agents with features like context compression and subagent spawning. LangSmith serves as the companion agent engineering platform, offering observability through tracing, evaluation (including LLM-as-judge and human feedback), deployment with scalable runtime, and Fleet for managing company-wide agents via natural language descriptions.

The system supports any LLM provider, includes SDKs for Python, TypeScript, Go, and Java, and integrates with OpenTelemetry for broader observability. It is used across the full development lifecycle from prototyping to production, with a focus on reliability, debugging, and scaling AI agents.