SYSTEM Cited by 1 source
LangGraph¶
LangGraph is a graph-based agent-orchestration framework from the LangChain ecosystem. It models an agentic workflow as a directed graph of nodes (agent / tool calls) and edges (state transitions), with a shared mutable state object passed between nodes. Used to implement supervisor-routing and multi-agent collaboration shapes that exceed the linear chain abstraction of plain LangChain.
Stub page. First wiki disclosure 2026-05-20 via the Databricks + Virtue Foundation post.
Why graph and not chain¶
LangChain's original chain abstraction is linear; LangGraph is a graph because realistic agent workflows have:
- Conditional branching — a supervisor classifies the input and routes to one of several specialist sub-agents.
- Cycles — a verification step may loop back to a re-planning step until a quality bar is met.
- Shared mutable state — tool outputs and intermediate reasoning need to flow between nodes without being re-derived.
LangGraph encodes these as a graph spec; the runtime executes the graph step-by-step, calling LLM endpoints / tools at each node.
Seen in¶
- sources/2026-05-20-databricks-virtue-foundation-medical-volunteers-72-countries — First wiki canonical source. Virtue Foundation's VF Agent prototype is built in LangGraph; the four-sub-agent composition (Medical Specialty Extractor + Multi-Agent Supervisor + Vector Search Agent + Genie Agent) is encoded as a LangGraph graph with the supervisor as the routing node.
Related¶
- systems/langchain — the parent framework.
- systems/vf-agent — first wiki canonical instance.
- patterns/multi-agent-supervisor-routing — the central pattern LangGraph supervisor-and-sub-agent compositions implement.