Your learning guide will appear here.
Back to course
Phase 2 · Orchestration & Protocols
LangGraph tutorial: build stateful agent workflows that survive the real world
Phase 2 of the Agentic AI Engineering course turns isolated model calls into orchestrated systems. Two of its eight modules are dedicated LangGraph deep-dives — this page maps what you will learn, and the AI guide generator below lets you practice any topic right now.
- Phase 2 modules
- 8
- Phase 2 modules
- LangGraph deep-dives
- 2
- LangGraph deep-dives
- Core graph skills
- 6
- Core graph skills
- Orchestrated agent you finish with
- 1
- Orchestrated agent you finish with
Why LangGraph
From one-shot calls to graphs that think in loops
Chains run once and stop. Real agents need to branch, retry, wait for a human, and pick up where they left off. LangGraph models that behaviour as a stateful graph — and Phase 2 teaches you to design, route, and persist those graphs before you ever touch production.
Covered by modules 3 and 4 of Phase 2
- Model agent state as a typed graph instead of a single prompt chain
- Route between nodes with conditional edges and deterministic control flow
- Add cycles so agents can retry, reflect, and loop until a goal is met
- Pause for human review with interrupts, then resume from a checkpoint
- Persist runs with checkpointers so long workflows survive restarts
- Compose sub-graphs and fan-out/aggregation for multi-step systems
The Phase 2 module map
Eight modules, two of them pure LangGraph
LangGraph sits in the middle of a wider orchestration toolkit. Each module below links back to its place in the full course curriculum — and the LangGraph modules can jump you straight into a generated practice guide.
- MODULE 01
LangChain Core — Chains, Memory & RAG
Compose reusable chains, memory strategies, retrievers, and grounded generation pipelines.
- MODULE 02
LangChain Agents & Tool Use
Build ReAct agents, typed tools, SQL integrations, retry logic, and streamed agent steps.
- MODULE 03LangGraph
LangGraph — Stateful Workflows & Routing
Model agent state, conditional routes, nodes, edges, and deterministic control flow.
- MODULE 04LangGraph
LangGraph — Cycles, Human-in-the-Loop & Persistence
Add review gates, checkpoints, resumability, fan-out, aggregation, and sub-graphs.
- MODULE 05
Tracing, Evaluation & Testing for LLM Applications
Trace runs, design datasets, measure quality, and prevent regressions.
- MODULE 06
Model Context Protocol — Architecture & Custom Servers
Build MCP tools, resources, prompts, transports, validation, and access controls.
- MODULE 07
MCP — Ecosystem Integrations
Connect agents to practical MCP services and manage multi-server environments.
- MODULE 08
Programmatic Prompting
Create declarative prompt programs, optimisers, metrics, and self-improving pipelines.
Practice now
Generate a course-aligned LangGraph guide
Choose a LangGraph topic and your level. Our AI-powered tutor builds a practical guide with code, three hands-on exercises, and how it maps to the Agentic AI Engineering course.
Ready to orchestrate agents for real?
Phase 2 is one of three phases in the Agentic AI Engineering course — 25 practical modules ending in a production capstone.