What is Learn LangGraph?
Learn LangGraph is an interactive course for LangGraph — the runtime for agents that need to loop, branch, remember, and pause. Not a new model. A graph over shared state.
You step a live graph in the browser and watch the jobs a chain cannot do: a ticket that looks again, a refund that waits, a researcher that isn’t done. The goal is to get someone from “I have heard of LangGraph” to “I can see what it is for, and I know when to leave the chain alone.”
Why We Built It
LangGraph is easy to name and hard to feel. The docs will give you nodes, edges, checkpointers, and interrupts. The thing that actually matters is the job: work that has to come back.
A list can only go forward. Support retries. Research re-queries. Refunds sit for a human. Twelve tickets become one standup brief. A flaky test gets edited until it passes. The user comes back on Tuesday. The moment you need a cycle, a pause, or a fan-out, you are already drawing a graph.
We built Learn LangGraph so people can step those jobs instead of only reading the vocabulary.
A Hands-On Way to See Why a List Dies
This is not a slide deck about agents.
You pick a module, press Step, and watch the token move. Lookup misses, the graph loops, state accumulates, an interrupt holds the run, a fan-out joins. No API key. Quizzes stay in this browser. Sign in with your En Dash account if you want progress and places on another machine.
Built Around Jobs, Then the Runtime
Two doors.
Use cases are the jobs: a ticket that looks again, a refund that waits, a question that isn’t done, twelve tickets for standup, tests until they pass, the same thread on Tuesday. Each one is “what a chain does” versus “what a graph does,” then a link into the matching module.
Learn is twelve modules on the LangGraph runtime itself: state and reducers, routing, ReAct cycles, checkpointers, interrupts, Send and Command, subgraphs, Store versus checkpointer, multi-agent. The capstone puts them in one graph.
Useful for Engineers Choosing a Shape
This is for people who already ship LLM features and are staring at a chain that keeps growing a special case.
Support and HITL. Research that has to re-query. Map-reduce over a pile of items. A coding agent that is just a loop. A run that has to survive until Thursday. If the path is A then B then C and it never comes back, stay on a chain. Learn LangGraph is for the other case.
Built to Make the Idea Click
Some product ideas fit in one sentence. LangGraph needs a loop you can watch.
Learn LangGraph exists so more people can see what the runtime is for, when a list is cheaper, and what it looks like when the work actually comes back.
Not official LangChain. A community resource from En Dash, same shop as html-in-canvas.dev and learn-pretext.com.