Deep Agents: From First Agent to Production Multi-Agent Systems

A hands-on workshop building real AI agent systems with LangChain’s Deep Agents framework.

What You’ll Learn

By the end of this workshop, you’ll be able to:

  • Master the four pillars of Deep Agents: LLMs, tools, planning, and memory — and understand how they work together to create autonomous agent systems

  • Build multi-agent systems with subagents: compose complex workflows by delegating specialized tasks to purpose-built subagents

  • Create and deploy reusable skills: package agent capabilities into shareable, configurable components that accelerate development

  • Ship production-ready agent systems: use the Deep Agents CLI to scaffold, test, and deploy agent applications

  • Apply your knowledge in a real-world capstone: build an AIOps agent system that monitors, diagnoses, and responds to infrastructure issues autonomously

How This Workshop Works

This workshop takes a top-down approach: you’ll start by running a working agent, then progressively peel back layers to understand how each component works. Each module is self-contained with hands-on exercises, and the final capstone brings everything together in a production-grade project.

Time commitment: Approximately 8 hours total, including exercises and the capstone project. You can work through modules at your own pace.

Learning approach: Each module follows a "working code first" philosophy. You’ll interact with functioning agents before diving into implementation details, making concepts concrete before exploring theory.

Prerequisites

To get the most out of this workshop, you should have:

  • Python experience: comfortable reading and writing Python code, understanding classes, decorators, and async/await patterns

  • Familiarity with LLM APIs: experience using OpenAI or Anthropic SDKs to call language models (even basic prompt/response patterns)

  • No LangChain or LangGraph experience required: we’ll introduce these frameworks from the ground up, focusing specifically on Deep Agents patterns

You’ll also need API keys for OpenAI or Anthropic to run the exercises.

Let’s Get Started

Ready to build your first AI agent system? Head over to the Workshop Overview to get oriented and set up your development environment.