Installation & First Graph

BeginnerTopic 2 of 22Open in Colab

Installation & First Graph

Get LangGraph installed and build your first stateful graph workflow.

Prerequisites

Python 3.9 or later
An OpenAI API key (or another LLM provider)

Install LangGraph

pip install langgraph langchain-openai

Set Up Your API Key

export OPENAI_API_KEY="sk-your-key-here"

Your First Graph

Let's build a simple graph that takes a topic, generates a joke about it, then evaluates whether the joke is funny.

from typing import TypedDict
from langgraph.graph import StateGraph, START, END
from langchain_openai import ChatOpenAI
 
llm = ChatOpenAI(model="gpt-4o-mini")
 
class State(TypedDict):
    topic: str
    joke: str
    rating: str
 
def generate_joke(state: State) -> dict:
    response = llm.invoke(f"Tell a short joke about: {state['topic']}")
    return {"joke": response.content}
 
def rate_joke(state: State) -> dict:
    response = llm.invoke(
        f"Rate this joke from 1-10 and explain why in one sentence:\n{state['joke']}"
    )
    return {"rating": response.content}
 
graph = StateGraph(State)
 
graph.add_node("generate", generate_joke)
graph.add_node("rate", rate_joke)
 
graph.add_edge(START, "generate")
graph.add_edge("generate", "rate")
graph.add_edge("rate", END)
 
app = graph.compile()
 
result = app.invoke({"topic": "AI agents"})
print(f"Joke: {result['joke']}")
print(f"Rating: {result['rating']}")

What Just Happened?

State — we defined a TypedDict with topic, joke, and rating fields
Nodes — two functions that take state, call the LLM, and return partial state updates
Edges — connected START → generate → rate → END
Compile & Run — graph.compile() produces a runnable app

The state accumulates data as it flows through each node. Each node only returns the fields it wants to update.

Visualizing the Graph

from IPython.display import Image
Image(app.get_graph().draw_mermaid_png())

This renders a visual diagram of your workflow — extremely useful for debugging complex graphs.

Next Steps

Now let's dive deeper into how state, nodes, and edges work together.

What is LangGraph?

State, Nodes & Edges