Skip to Content
View all events

Building AI Agents and Workflows with LangGraph

Published by Pearson

Intermediate content levelIntermediate

From prompts to structured, stateful systems

  • Learn to move from isolated prompts to real AI workflows using LangGraph with memory, tools, control flow, and recoverable state.
  • Build working agents step-by-step including tool-using agents with MCP, evaluators, and multi-step workflows you can actually reuse in production.
  • Understand where LangChain alone stops being enough and how LangGraph solves orchestration, state, and reliability for agentic systems.

This class is a practical, code-first guide to building real AI agents and workflows using LangGraph. Instead of isolated prompt engineering, we focus on designing systems that have memory, use tools, integrate RAG pipelines, recover from errors, branch logic, and maintain state across steps. You will learn how LangGraph extends LangChain by solving orchestration, control flow, and observability, and how to use it to build agents that are reliable, testable, and deployable.

As AI moves from experiments to real products, teams need more than clever prompts. They need workflows they can debug, RAG systems that do not hallucinate, and agents that reason across multiple steps with context. This course shows you how to make that jump. Through examples and live coding, you will learn how to model tasks as graphs, connect tools and retrieval systems, add human or automated evaluators, and ship agent systems that are stable and understandable.

What you’ll learn and how you can apply it

  • Build stateful AI agents using LangGraph with tools, memory, RAG components, and controlled workflow execution.
  • Design and deploy repeatable multi-step workflows that handle branching, retries, human in the loop, and error recovery instead of brittle prompt chains.
  • Integrate retrieval systems and vector databases into your agents using LangChain and LangGraph to build grounded, context-aware RAG pipelines.
  • Evaluate and refine agent and RAG performance with automated checks, human-in-the-loop review, and iterative graph improvements.

This live event is for you because...

  • You are a software developer, data scientist, or ML engineer who wants to move beyond single prompts and build real AI agents and workflows.
  • You are already using or exploring LangChain or RAG and want to understand when LangGraph becomes the better tool for structure, reliability, and state management.
  • You are building AI-powered products and need agents that can use tools, retrieve knowledge, reason across steps, and be debugged and deployed with confidence.

Prerequisites

  • Basic to intermediate Python experience. You should be comfortable reading and writing Python scripts.
  • Familiarity with large language models and prompting. You should know how to call an LLM through an API or use LangChain at a basic level.
  • Introductory knowledge of RAG or embeddings. This is helpful for understanding how retrieval fits into agent workflows but not strictly required.
  • No prior LangGraph experience required. We will start from first principles.

Course Set-up

  • Python environment with Python 3.9 or later installed.
  • GitHub repository access. All course materials, notebooks, and example workflows will be provided in a GitHub repo before the session. https://github.com/sinanuozdemir/oreilly-langgraph
  • Required libraries installed. Instructions and a requirements.txt file will be included in the repository. This will cover packages like LangChain, LangGraph, OpenAI or Anthropic SDKs.
  • API keys for at least one LLM provider (OpenRouter, OpenAI, Anthropic, or similar). Optional but recommended for hands-on testing with real models.

Recommended Preparation

Recommended Follow-up

Schedule

The time frames are only estimates and may vary according to how the class is progressing.

Segment 1: From Prompts to Workflows (25 minutes)

  • When single prompts are not enough
  • What LangGraph solves compared to LangChain alone
  • Overview of agents, workflows, and AI state management

Segment 2: LangGraph Basics (35 minutes)

  • Nodes, edges, and graph execution
  • Memory, state, and checkpoints
  • Exercise: Build a LangGraph AI agent

Q&A + Break (10 minutes)

Segment 3: Tools and RAG Integration (35 minutes)

  • Adding tools, APIs, and MCP functions
  • Connecting retrieval and vector databases into agent flows
  • Exercise: Build a RAG-enabled workflow and agent with LangGraph and MCP

Segment 4: Workflow Control and Error Recovery (30 minutes)

  • Branching logic and conditional steps
  • Handling failure, retries, and human-in-the-loop decisions
  • Exercise: Add error handling and tracing to an existing agent using Langsmith

Q&A + Break (10 minutes)

Segment 5: Evaluation and Iteration (40 minutes)

  • Testing agent behavior and debugging workflows
  • Automated evaluators and feedback loops
  • Exercise: Add evaluation to an agent workflow

Q&A + Break (5 minutes)

Segment 6: Deployable Agents and Real Use Cases (40 minutes)

  • Packaging LangGraph workflows for production
  • Architecture patterns for single-agent and multi-agent systems
  • Case studies from real systems including a multi-agent example

Segment 7: Wrap-up and Next Steps (10 minutes)

  • Key takeaways
  • Resources and follow-up learning paths
  • Final Q&A

Your Instructor

  • Sinan Ozdemir

    Sinan Ozdemir is the founder of Crucible, an AI factory platform that helps teams convert existing workflows into custom models. He is a Y Combinator alum, AI & LLM Advisor at Tola Capital, and the author of multiple books on data science and machine learning including Building Agentic AI, Quick Start Guide to LLMs, and Principles of Data Science. Sinan is a former lecturer of data science at Johns Hopkins University and the founder of Kylie.ai, an enterprise-grade conversational AI platform (acquired 2014). He holds a master's degree in pure mathematics from Johns Hopkins University and is based in San Francisco, California.

    linkedinXlinksearch

Skill covered

Generative AI