langgraph-fundamentals

Define a StateGraph with nodes and edges to control execution flow, then compile and invoke the graph with an input payload to execute multi-step AI tasks with state updates.

Use Command-based routing and edge types to direct execution flow in LangGraph agent workflows, enabling dynamic branching and sequential state updates across nodes.

Yes, you can run parallel tasks in a LangGraph StateGraph by using the Send API to dispatch multiple operations concurrently, enabling parallel sends within your directed graph workflows.

Stream state updates from a LangGraph runtime by using the stream API instead of invoke, allowing you to monitor execution flow and surface real-time UX updates during multi-step agent workflows.

Yes, reducers are required to manage state updates in LangGraph workflows, defining how state transitions occur across nodes and ensuring accurate data flow throughout the directed graph execution.

Use LangGraph for orchestrating AI tasks when you need complex decision flows, human-in-the-loop prompts, or branching logic, as it models these operations as a structured directed graph with clear error handling.