LangGraph is a new package for creating LLM workflows containing cycles, which are crucial components of most agent runtimes. Today, we're highlighting multi-agent workflows using LangGraph. Multi-agent designs allow dividing complicated problems into tractable units of work that can be targeted by specialized agents and LLM programs. We've added three separate examples of multi-agent workflows to the langgraph repo: Multi Agent Collaboration, Agent Supervisor, and Hierarchical Agent Teams. Additionally, we're excited to highlight a few applications built on top of LangGraph that utilize the concept of multiple agents, such as GPT-Newspaper and CrewAI.