LangChainLangGraph is a low-level orchestration framework for building, managing, and deploying long-running, stateful agents. Rather than prescribing an agent architecture, it supplies supporting infrastructure for any stateful workflow, and it can be used standalone or with the wider LangChain ecosystem: Deep Agents is a higher-level package built on top of it for planning, subagents, and file-system use, LangChain provides integrations and components, and LangSmith covers debugging, evals, and deployment. A JS/TS equivalent, LangGraph.js, is maintained in a separate repository.
The design draws on Pregel and Apache Beam, and the public interface takes inspiration from NetworkX. LangGraph is built by LangChain Inc but does not require LangChain.
Features
- Durable execution: agents persist through failures and resume from exactly where they left off
- Human-in-the-loop: interrupts let a person inspect and modify agent state at any point during execution
- Memory: short-term working memory within a run and long-term memory across sessions
- Debugging with LangSmith: visualization of execution paths, state transitions, and runtime metrics
- Deployment: LangSmith Deployment runs long-running stateful workflows, with visual prototyping in LangSmith Studio
- Ecosystem fit: Deep Agents, LangChain integrations, and LangSmith observability plug in without changing the graph model
