Mark Cavage, Tushar Jain, Jim Clark & Yunong Xiao

Compose the Future: Building Agentic Applications, Made Simple with Docker

What if you could build and deploy a multi-agent AI system using just one file? See how Docker Compose makes complex agentic applications simple.

Compose the Future: Building Agentic Applications, Made Simple with Docker
#1about 2 minutes

Why Docker is simplifying agentic application development

Docker's role in agentic development is analogous to how it made microservices accessible by simplifying complex environments.

#2about 4 minutes

Defining agentic applications and their core properties

An agent is an autonomous system where the LLM drives the control loop, connects to your data, and operates based on goals rather than tasks.

#3about 3 minutes

The future of multi-LLM and compound agents

Future agents will be multi-LLM and compound, requiring solutions for model portability, tool discovery via MCP, and a trusted hub for components.

#4about 6 minutes

Introducing Docker's new tools for building agents

Docker provides key building blocks for agents, including the Docker Model Runner for local models and the MCP Gateway for secure tool integration.

#5about 2 minutes

Extending Docker Compose with native agent support

Docker Compose now natively supports agent development with top-level keywords for models and tools, simplifying the entire agent stack definition.

#6about 8 minutes

Live demo of building a multi-agent e-commerce application

A practical demonstration shows how to build a sock store agent using Docker Compose to manage cooperating agents, hybrid LLMs, and an MCP gateway.

#7about 7 minutes

Accelerating local development with Docker Offload

Docker Offload allows developers to seamlessly switch their local Docker context to a remote, powerful GPU in the cloud to accelerate model-heavy workloads.

#8about 13 minutes

Deploying agentic applications with serverless compose on GCP

A partnership with Google Cloud enables deploying the same Docker Compose file directly to GCP Cloud Run for serverless, scalable production environments.

#9about 3 minutes

Introducing the Docker Agent for automating the SDLC

Docker is becoming an agent itself, capable of automating tasks like finding and fixing security vulnerabilities and optimizing Dockerfiles.

#10about 2 minutes

Recap of new tools and special offers

A summary of the new agentic platform capabilities, including Compose extensions, Docker Offload, and serverless deployment, is presented with a special offer.

Related jobs
Jobs that call for the skills explored in this talk.

Featured Partners

From learning to earning

Jobs that call for the skills explored in this talk.

DevOps Engineer (f/m/d)

DevOps Engineer (f/m/d)

Power Plus Communications
Mannheim, Germany

Intermediate
Senior
GIT
Linux
Docker
Kubernetes