Share feedback
Answers are generated based on the documentation.

Lab: Building Agentic Apps with Docker

Get up and running with building agentic applications using Compose, Docker Model Runner, and the Docker MCP Gateway. This hands-on lab takes you from understanding AI models to building complete agentic applications.

What you'll learn

This lab covers three core areas of agentic application development:

Models: What models are, how to interact with them, configuring Docker Model Runner in Compose, and writing code that connects to the Model Runner

Tools: Understanding tools and how they work, how MCP (Model Context Protocol) fits in, configuring the Docker MCP Gateway, and connecting to the MCP Gateway in code

Code: What agentic frameworks are, defining models and tools in a Compose file, and configuring your app to use those models and tools

Modules

#ModuleDescription
1IntroductionOverview of agentic applications and the Docker AI stack
2Understanding Model InteractionsLearn how to interact with AI models
3The Docker Model RunnerConfigure and use Docker Model Runner with Compose
4Understanding Tools and MCPDeep dive into tools, tool calling, and MCP
5The Docker MCP GatewaySet up and configure the MCP Gateway
6Putting It All TogetherBuild a complete agentic application
7ConclusionSummary and next steps

Prerequisites

  • Install the latest version of Docker Desktop
  • Enable Docker Model Runner by going into Settings in Docker Desktop, choosing AI, then selecting Docker Model Runner
  • Pull the Gemma 3 model before launching by running this command:
$ docker model pull ai/gemma3

Launch the lab

Start the labspace:

$ docker compose -f oci://dockersamples/labspace-agentic-apps-with-docker up -d

Then open your browser to http://localhost:3030.

Note

It may take a little while to start due to the AI model download.