Docker Compose your Dev Toolkit for AI and Cloud workflows

Docker Compose has always been a developer favorite for orchestrating local environments, but it just got a major upgrade for the AI era.
In this fully live demo session (no fluff, almost no slides), I’ll showcase how the latest features in Docker Compose like the new models attribute and provider services, make it possible to declaratively define and run LLM-powered applications with ease.

You'll see how to plug an AI model into your app using the models section, connect to a remote by using a Telepresence provider service, and when your laptop can’t keep up? Docker Offload will seamlessly run your heavy workloads remotely, without changing your dev flow.
What you’ll see live:

• Defining an LLM (like llama3, qwen, etc.) in a Compose file using the models key
• How you can easily access to those LLM from your service containers
• The power of provider services to prepare or connect to Compose external resources
• Running GPU-heavy tasks remotely via Docker Offload, when local compute isn’t enough
• Using Compose as the single source of truth to orchestrate everything from code to context

This session is fast, practical, and built for developers working with AI, agents, or modern multi-service stacks.
You’ll walk away with a concrete understanding of how Docker Compose’s latest features help you develop AI applications