Graph-based framework for building stateful, controllable multi-agent workflows.
Category: Agent Frameworks · Type: Agent orchestration framework · Website: https://www.langchain.com/langgraph
Overview
LangGraph, from the LangChain team, models agents as graphs of nodes and edges with explicit state. This gives developers fine-grained control over loops, branching, human-in-the-loop checkpoints, and durable execution — addressing the reliability gaps of free-form agent loops.
Key Features
- Explicit graph/state machine model for agents
- Durable execution, checkpointing, and human-in-the-loop
- First-class multi-agent patterns
- Pairs with LangSmith for tracing and LangChain components
Role in the Agentic AI Stack
A production-grade orchestration layer that calls foundation models and tools; debugged and monitored with LangSmith.
Part of Agentic AI Landscape → Agent Frameworks