A tool for making apps that can remember and learn from conversations, using graphs to manage interactions. It’s built on LangChain to help create smarter, more interactive apps.
In the LangGraph framework, cognitive architectures are the basis, they are designed to enable cycles in runtime, allowing agents to use reasoning tasks effectively.
Cognitive architectures visualisation:
Key components of LangGraph:
- LangGraph: A module for creating cyclical graphs in LangChain.
- Cyclical Graphs: Enable loops in chains for complex applications.
- StateGraph: A class for graph representation with updateable state.
- Nodes: Functional units in the graph, can be functions or runnables.
- Edges: Connections between nodes for flow control.
- AgentExecutor: A specific implementation for agent-like behavior.
- Chat Agent Runtimes: Adapted for chat models and message lists.
- Modifications: Customizations like tool forcing and output formatting.
- Future Enhancements: Advanced runtimes, stateful tools, and multi-agent workflows.
TL;DR: LangGraph is module built on top of LangChain to better enable creation of cyclical graphs, often needed for agent runtimes. * Python Repo * Python YouTube Playlist * JS Repo Introduction One of the things we highlighted in our LangChain v0.1 announcement was the introduction of a new library: LangGraph.