LangGraph

What it is

LangGraph is a graph-based framework for building stateful, multi-actor LLM applications. You define agents as nodes and edges, with explicit state and conditional branching — closer to a workflow engine than a black-box agent loop.

Why people use it

• Explicit control over agent state and transitions (good for complex, auditable flows).
• Strong LangChain ecosystem integration.
• Checkpointing and human-in-the-loop built in.

Watch-outs

• More boilerplate than lighter harnesses; the graph mental model has a learning curve.
• Tight coupling to the LangChain stack can be a plus or a lock-in depending on your view.