Loading...
Loading...
Stop writing agents that forget everything between turns. Build a LangGraph state machine with a shared TypedDict, an LLM router, slot-collection nodes, a catalog lookup, a propose-and-confirm flow, and MemorySaver checkpoints. Every turn resumes exactly where the last one left off.
Message a mentor about fit, prerequisites, or where to start. Replies come on WhatsApp, usually within a day.
Engineers are learning here from
Build a long-running, multi-turn agent as a LangGraph state machine. Share typed state across intent routing, slot collection, catalog lookup, option proposal, and confirmation nodes, then checkpoint every turn with MemorySaver so conversations resume exactly where they left off.
Build a long-running, multi-turn agent with typed state, intent routing, and checkpointed memory.
What you'll ship
What you'll learn
Curriculum
Typed agent state
Define the shared TypedDict that every node in the graph reads from and writes to
Intent router
Build an LLM-powered node that classifies every turn and extracts slots in one call
Slot-collection nodes
Build focused nodes that each ask one targeted question and validate the answer
Catalog lookup node
Query a static dataset, filter the results, and hand the top candidates back to the graph
Propose and confirm
Capture the user choice with an LLM fallback and write a confirmation record
MemorySaver persistence
Checkpoint every turn with MemorySaver so conversations resume across requests
Who it's for
who have written single-shot LLM calls but want their agents to hold state across turns
who need to add stateful conversational flows to an existing API
who want to understand how LangGraph nodes, edges, and checkpointers actually work under the hood
FAQ
No. The course starts from a bare StateGraph and builds up one node at a time. If you have called an LLM from Python before, you have enough background.
The workshop ships with a provider pattern that supports OpenRouter, Fireworks, Gemini, and OpenAI. You only need one API key, and swapping providers does not touch the graph code.
Flights are the demo domain, but the techniques apply to any stateful agent: support triage, onboarding flows, order management, scheduling. The patterns are about stateful orchestration, not travel.
Pricing
Subscribe to Pro for every paid course, or buy just this one.
Unlock this course and every paid course plus workshop replays. One subscription.
You save 54% with regional pricing
One-time purchase. Lifetime access to every lesson, exercise, and update.
You save 47% with regional pricing
Still deciding? Ask Param a question
Stateful agent workflows with LangGraph
$79 one-time