In the rapidly evolving landscape of technology, the concept of "agentic applications" is becoming increasingly relevant. These applications, defined by their ability to utilize Large Language Models (LLMs) to dictate execution workflows based on specific goals while interacting with various tools, data, and systems, are poised to transform the way we approach software development. At the WeAreDevelopers conference, significant strides were made in simplifying the creation, testing, and deployment of these applications, with Docker at the forefront of these innovations.
Understanding Agentic Applications
Agentic applications are a new breed of software that leverage advanced AI models to perform complex tasks autonomously. They integrate seamlessly with existing tools and systems, creating a cohesive environment where AI-driven processes can operate efficiently. This new application stack, while innovative, shares similarities with microservices, a more established technology that also focuses on modularity and scalability. However, agentic applications require a different approach to tooling and development processes due to their reliance on AI models.
Docker’s Role in Evolving Agentic Applications
Docker has been instrumental in the evolution of agentic applications. Over recent months, the company has been refining its tools to accommodate the unique requirements of this new application stack. At the WeAreDevelopers conference, Docker unveiled several key initiatives designed to enhance the development and deployment of agentic applications.
Key Announcements from WeAreDevelopers
Docker’s President and COO, Mark Cavage, alongside EVP of Product and Engineering, Tushar Jain, presented a keynote at the WeAreDevelopers conference, introducing groundbreaking features such as Compose for agentic applications, native support for Google Cloud, and Docker Offload. These announcements mark significant advancements in the ease of building and deploying agentic applications.
Compose for Agentic Applications
Compose, a tool traditionally used for defining and running multi-container Docker applications, has been enhanced to support the unique needs of agentic applications. These applications require models, tools, and custom code to work in harmony. Docker Model Runner allows developers to download and run models, while the open-sourced MCP Gateway offers a secure environment for running containerized MCP servers. With Compose, developers can now define and connect these elements within a single YAML file, streamlining the development process.
For instance, a typical Compose file for an agentic application might specify models such as "gemma3" and services like the "mcp-gateway" that provide necessary tools. This integration facilitates a seamless connection between models, tools, and the application itself, enabling efficient execution of tasks.
Cloud-Provider Support for Production Deployment
During the keynote, Docker showcased a collaboration with Google Cloud, demonstrating how Compose-based applications can be easily deployed using Google Cloud Run. This capability ensures that the same Compose configuration can be used from development through to production without requiring rewrites or reconfigurations. The added support for model deployment further enhances the utility of Compose in cloud environments.
This integration with Google Cloud Run, and soon Microsoft Azure Container Apps, allows developers to deploy applications to serverless platforms effortlessly. This consistency and portability make the transition from development to production smooth and efficient.
Introducing Docker Offload
Running LLMs typically demands substantial computational resources, often beyond the reach of individual developers. Docker Offload addresses this challenge by allowing developers to run containers and models using cloud-based compute resources. This feature provides the necessary GPU power without the complexities of local setup or configuration issues. Docker Offload is as simple as a toggle switch in Docker Desktop, offering 300 free GPU minutes for early adopters who sign up for the beta program.
Hands-On Experience with New Tools
The WeAreDevelopers conference also featured a hands-on workshop where attendees could explore Docker’s new Compose capabilities. This workshop provided a practical understanding of the agentic application stack, focusing on models, tools, and agentic frameworks. Participants used Docker Model Runner, the Docker MCP Gateway, and Compose integrations to package and deploy applications effectively.
The workshop’s popularity was evident as every seat was filled, with additional attendees standing by, eager to participate. The engaging session underscored the growing interest in agentic applications and the practical benefits of Docker’s new tools.
Inspirational Lightning Talks
In addition to the workshop, Docker hosted a series of lightning talks at its booth, sparking new ideas and discussions around agentic applications. These talks covered topics such as:
- The fundamentals of GenAI applications
- The potential of event-driven agentic applications beyond chatbots
- Using LLMs for semantic testing of applications and websites
- Building safer and more secure images with Docker Hardened Images
These sessions highlighted the practical applications of agentic apps and emphasized Docker’s central role in their development.
Industry Insights and Future Plans
Docker’s UX Research team shared insights from three years of industry research, revealing how the AI landscape is influencing language selection, security trends, and more. This presentation provided valuable context for understanding the broader impact of AI on software development.
Looking ahead, Docker announced a partnership with WeAreDevelopers to launch the WeAreDevelopers World Congress North America in September 2026. This event promises to bring the dynamic and engaging atmosphere of the European conference to a new audience, offering developers a platform to explore cutting-edge technologies and build their careers.
Conclusion
The advancements in agentic applications showcased at the WeAreDevelopers conference signal a new era in software development. Docker’s commitment to simplifying the creation and deployment of these applications through enhanced tools like Compose and Docker Offload is paving the way for more accessible and efficient AI-driven solutions. As the field continues to evolve, developers are encouraged to leverage these tools in their workflows, contributing to the growth and innovation of agentic applications. For more information, visit Docker’s official website.
For more Information, Refer to this article.
