Gen-AI Developer Classroom notes 06/Apr/2026

Langgraph

• Langgraph is a framework which helps in creating a controlled and reliable agent.
• The steps required to solve a problem will be defined by us as a workflow.
• Basic idea is to create a graph (workflow)

Example scenario

• Student Assignment Tracker
  

• Langgraph is perfect for these kind of scenarios

Langgraph Way of solving above problem

• Langgraph defines graph and refers it as StateGraph
• State is share data across nodes
• node is a computation unit (function)
• edge connects nodes

Lets create our first langgraph

• A state can be

  • TypedDict
  • Dataclass
  • Pydantic Model

• create a new folder hello-langgraph

• Initialize with uv and add langgraph pacakge

uv init
uv add langgraph langgraph-cli

• A node in a langraph will have state as input and can have following outputs
  • complete state
  • partial state
• Create a file called as hello.py with following code

"""Hello langgraph 
"""
from typing import TypedDict
from langgraph.graph import StateGraph, START, END

# lets create state
class MyState(TypedDict):
    """This class represents the state of the graph
    """
    name: str
    friends: list[str]
    family: list[str]

# node 1
def find_friends(state: MyState) -> MyState:
    """This function finds friends
    """
    state['friends'] = ['Ram', 'Shyam']
    return state

# node 2
def find_family(state: MyState) -> MyState:
    """This function finds family
    """
    state['family'] = ['sita', 'gita']
    return state

# graph which takes state as argument
graph = StateGraph(MyState)

graph.add_node("friends", find_friends)
graph.add_node("family", find_family)
graph.add_edge(START, "friends")
graph.add_edge("friends", "family")
graph.add_edge("family", END)

# compile the graph
compiled_graph = graph.compile()


if __name__ == "__main__":
    response = compiled_graph.invoke({
        "name": "abc"
    })
    print(response)

• Now create a file called as langgraph.json in the root of project with following values

{
  "dependencies": ["."],
  "graphs": {
    "hello": "hello:graph"
  },
  "env": ".env"
}

• If you want to run the graph uv run hello.py
• If you want to view in langgraph studio langgraph dev

• Ensure virtual enviornment is activated other wise use uv run langgraph dev

• Refer Here for changes