"Docker delves into the AI field, empowering developers to build GenAI applications"

In almost all generative AI applications used for training or inference today, you may find Docker containers as the primary deployment method.

At Dockercon in Los Angeles, Docker is working on a series of initiatives aimed at helping developers build generative AI applications faster.

One of these initiatives is the launch of a new product called GenAI Stack, which combines Docker with Neo4j graph database, LangChain model linking technology, and Ollama integration for running large language models (LLM). Additionally, the new Docker AI product will also make its debut at Dockercon as a developer tool that integrates AI-driven insights and container development directions.

The importance of Docker to the modern development ecosystem is undeniable, and the new AI initiatives could have a significant impact on the development work of GenAI. In recent years, Docker has increased its focus on developers, and the company's CEO says that this effort is paying off.

Docker CEO Scott Johnston said, "For four consecutive years, the Stack Overflow developer community has ranked us as the most popular and beloved development tool, and we now have 20 million monthly active developers from around the world."

What Docker GenAI Stack brings to developers

Although using Docker containers to aid in sharing and deploying AI is already common, Johnston states that simplifying the development of GenAI applications is still necessary.

Typically, GenAI applications require some core elements, such as a vector database, which Neo4j now possesses as part of its graph database platform. Additionally, GenAI also requires LLM, which is provided by Ollama, a platform that allows users to run LLMs, including Llama 2, locally. Modern GenAI applications are often multi-step, and this is where LangChain matches with its framework. Configuring all these different components in a container to make them work together usually requires some effort, but now this process can be greatly simplified through the GenAI Stack.

The Docker GenAI Stack aims to help developers and their organizations start AI development more easily using containers. With the GenAI Stack, developers can begin coding and experimenting to drive development forward as it is pre-configured and ready to use," Johnston said.

The entire stack is designed to run locally on developers' systems and is provided for free. When developers build applications and need deployment and commercial support, Docker and its partners will offer corresponding options, according to Johnston.

Docker AI: The "mech suit" for developers

There is no shortage of GenAI development tools in the market today, with popular choices including GitHub Copilot and Amazon CodeWhisper.

Docker is now entering this field with its own GenAI tool, known as Docker AI. Unlike Microsoft and other vendors increasingly referring to their GenAI tools that assist users as collaboration tools, Docker uses the term "mech suit." The basic idea is that developers, with the help of the mech suit, have greater power and capability to complete tasks.

Docker AI has been trained using Docker's proprietary data, including millions of Dockerfiles, compose files, and error logs. Docker AI is integrated directly into developers' workflows to provide assistance when errors occur. It displays potential fixes in the development environment and allows developers to test the proposed fixes before committing changes. The goal is to provide developers with a better troubleshooting and debugging experience when issues arise.

Johnston points out that while tools like GitHub Copilot are useful and powerful, Docker AI is specifically tailored for container development.

"It's trained on rich proprietary data from Docker that other LLMs don't have access to," he said.