LangGraph models an agent as a graph of steps with explicit, checkpointed state, which makes long-running and resumable workflows practical. Interrupts let a graph pause for human review before a step runs. Its persistence layer decides where conversation state, including sensitive data, is stored.
Read in this order
- OfficialLangGraph overview (LangChain Docs)Start here: the graph and state model that the rest of the framework builds on.
- OfficialInterrupts (LangGraph)How to stop a graph for human approval before a step runs.
- OfficialPersistence (LangGraph)Where state is saved, which tells you what sensitive data gets stored.