Welcome
Getting Set Up (Python)
Getting Set Up - Text (Python)
Getting Set Up (TypeScript)
Getting Set Up - Text (TypeScript)
Language Translation
Module 1: Introduction
Lesson 1: Deep Agents Overview
Lesson 2: Running a Deep Agent
Lesson 3: Models
Lesson 4: The System Prompt
Lesson 5: Tools
Lesson 6: MCP
Lesson 7: Messages, Threads, and Checkpointers
Lesson 8: HITL
Test Your Skills
Module 1 Feedback
Module 2: Introduction
Lesson 1: The Deep Agent Environment
Lesson 2: The Filesystem Backends
Lesson 3: Sandboxes and LocalShell
Lesson 4: Interpreter
Module 2 Feedback
Module 3: Introduction
Module 3: Overview
Lesson 1: Summarization and Offload
Lesson 2: Skills
Lesson 3: Memory
Module 3 Feedback
Module 4: Introduction
Lesson 1: Delegation
Lesson 2: Building a Subagent Team
Lesson 3: Dynamic Subagents
Module 4 Feedback
Module 5: Introduction
Lesson 1: Project Overview
Lesson 2: Local Deployment
Lesson 3: The Sales Assistant
Lesson 4: Asynchronous Subagents
Lesson 5: The Advanced Sales Assistant
Module 5 Feedback
Choose Deep Agents if you want an agent that works on long-running, complex tasks out-of-the-box. Note: Deep Agents are actually built on top of LangChain building blocks and executed using the LangGraph runtime!
Deep Agents can be used to build deep research, multi-file coding, or any kind of autonomous assistant workflows where the agent needs to plan, use a filesystem, and delegate on its own.
The term "deep agents" is used in contrast to "shallow agents" to describe AI agents that can tackle complex, open-ended, and long-horizon tasks rather than just simple, single-step prompts.