Skip to main content
A LangGraph workflow is built using several core elements. Continuing our Restaurant Kitchen analogy, let’s explore how these elements represent code in practice.

​
1. State 🧮

The State is a shared data structure that holds the current information or context of the entire application. It acts as the application’s memory.
  • Code representation: Usually defined as a Python TypedDict or a Pydantic model.
  • Analogy: The Order Tray that holds the ticket and the plate of food as it travels through the kitchen.

​
2. Nodes 📌

Nodes are individual functions or operations that perform specific tasks within the graph. They receive the current state, do some processing, and return an updated dictionary to modify the state.
  • Code representation: Python functions or LangChain Runnables.
  • Analogy: The Stations (Prep station, Grill station) where cooks perform tasks on the plate.

​
3. Graph 🍡

The Graph is the overarching structure that maps out how different tasks (nodes) are connected and executed.
  • Code representation: Configured using the StateGraph class.
  • Analogy: The Kitchen Blueprint showing where all stations are located and how trays move between them.

​
4. Edges 🔥

Edges are the connections between nodes that determine the flow of execution. They tell the graph which node to go to next.
  • Code representation: Added using graph.add_edge(source_node, target_node).
  • Analogy: The Conveyor Belt or the rule directing a tray from Prep Station straight to Grill Station.

​
5. Conditional Edges 🎏

Conditional Edges are specialized connections that decide the next node to execute based on specific conditions or logic applied to the current state.
  • Code representation: Added using graph.add_conditional_edges(), pointing to a routing function.
  • Analogy: The Quality Control Inspector who checks if the food is cooked correctly and routes the tray accordingly.

​
6. START 🚀 & END 🚩

These are special virtual nodes that define where the graph begins and where it finishes execution.
  • Code representation: Imported as START and END from langgraph.graph.
  • Analogy:
    • START: The Order slip printer printing a new customer order.
    • END: The Service Window where the waiter picks up the finished dish.

​
7. Tools 🛠 & ToolNode 🔎

  • Tools: Executable utilities that nodes can call (like web search, calculations, or API requests).
  • ToolNode: A specialized node in LangGraph designed to execute tools automatically and write their outputs back to the State.
  • Analogy:
    • Tools: The kitchen appliances (microwave, blender, meat thermometer).
    • ToolNode: The assistant cook whose sole job is to operate the blender when requested by the Chef.

​
8. Messages 📬

In conversational agents, the State often contains a list of messages representing the chat history. LangGraph uses standard message classes: