How to beginner · 3 min read

How to create a graph in LangGraph

Q: How to create a graph in LangGraph

To create a graph in LangGraph, instantiate a Graph object, add nodes and edges using add_node() and add_edge() methods, then visualize or query the graph. Use LangChain to integrate AI capabilities for graph-based reasoning and data linking.

Quick answer

To create a graph in LangGraph, instantiate a Graph object, add nodes and edges using add_node() and add_edge() methods, then visualize or query the graph. Use LangChain to integrate AI capabilities for graph-based reasoning and data linking.

PREREQUISITES

Python 3.8+
pip install langchain langgraph
OpenAI API key (free tier works)
Set environment variable OPENAI_API_KEY

Setup

Install the required packages and set your OpenAI API key in the environment variables.

bash

pip install langchain langgraph

Step by step

This example shows how to create a simple graph, add nodes and edges, and print the graph structure using LangGraph.

python

from langgraph import Graph
import os

# Create a graph instance
graph = Graph()

# Add nodes
node1 = graph.add_node('Node1', data={'description': 'Start node'})
node2 = graph.add_node('Node2', data={'description': 'Second node'})
node3 = graph.add_node('Node3', data={'description': 'Third node'})

# Add edges
graph.add_edge(node1, node2, label='connects_to')
graph.add_edge(node2, node3, label='leads_to')

# Print graph nodes and edges
print('Nodes:')
for node in graph.nodes:
    print(f"{node.id}: {node.data}")

print('\nEdges:')
for edge in graph.edges:
    print(f"{edge.source.id} -[{edge.label}]-> {edge.target.id}")

output

Nodes:
Node1: {'description': 'Start node'}
Node2: {'description': 'Second node'}
Node3: {'description': 'Third node'}

Edges:
Node1 -[connects_to]-> Node2
Node2 -[leads_to]-> Node3

Common variations

You can create directed or undirected graphs, add metadata to nodes and edges, and integrate AI models from LangChain for advanced graph reasoning. Async graph operations and streaming updates are also supported in newer versions.

python

from langgraph import Graph
from langchain_openai import ChatOpenAI
import os

# Initialize graph and AI client
graph = Graph()
client = ChatOpenAI(model='gpt-4o', openai_api_key=os.environ['OPENAI_API_KEY'])

# Add nodes
nodeA = graph.add_node('A')
nodeB = graph.add_node('B')

# Add edge
graph.add_edge(nodeA, nodeB, label='related_to')

# Use AI to generate edge label dynamically
prompt = f"Suggest a relationship label between {nodeA.id} and {nodeB.id}."
response = client.chat([{"role": "user", "content": prompt}])
label = response.content.strip()

# Update edge label
graph.edges[0].label = label

print(f"Edge label updated to: {label}")

output

Edge label updated to: influences

Troubleshooting

If you see ModuleNotFoundError, ensure langgraph is installed via pip install langgraph.
If environment variables are not set, set OPENAI_API_KEY before running your script.
For graph visualization issues, check if your environment supports graphical output or export the graph data for external visualization tools.

✅

Key Takeaways

Use Graph from langgraph to create and manage graph nodes and edges easily.
Integrate LangChain AI models to dynamically generate or analyze graph relationships.
Always set your API keys securely via environment variables for production-ready code.

Verified 2026-04 · gpt-4o

Verify ↗