 |
VOOZH | about |
|
VOOZH | about |
We’re so glad you’re here. You can expect all the best TNS content to arrive Monday through Friday to keep you on top of the news and at the top of your game.
Check your inbox for a confirmation email where you can adjust your preferences and even join additional groups.
Follow TNS on your favorite social media networks.
Become a TNS follower on LinkedIn.
Check out the latest featured and trending stories while you wait for your first TNS newsletter.
Docker unveiled two new artificial intelligence solutions at last week’s DockerCon: A generative AI stack to help developers get started with creating their own AI applications, and Docker AI, which is designed to help developers deploy and optimize Docker itself.
It’s the first AI offering for a product that is itself often used to build popular AI tools, such as Hugging Face and Replicate. Many expose the interfaces to their communities as a Docker container and ask their communities to submit models to their services as Docker container images, according to Docker CEO Scott Johnston.
“Container images are what’s called immutable, meaning they’re reproducible,” Johnston told the New Stack. “When you’re doing modeling, science and scientific analysis, you want to be able to reproduce the results regardless of the colleague, regardless of the situation. So the fact that the container image is immutable really helps with that.”
Docker has worked with these upstream communities to provide trusted images for AI/ML activities and applications for years, he added.
“With the rise in interest in this category, the last couple of years we’ve seen the downloads just spike through the roof,” he said. “We’re seeing the community download those, make their own tweaks, tune, and then package them up again as Docker containers and share them with each other through Hub.”
Hub is Docker’s public registry service for Docker images. It has more than 500 AI images now being shared through the Hub, he added.
Docker’s community has asked how they can get started with AI/ML technologies, Johnston said.
“We hear a lot from our developer community, ‘How do I get started? What tech can I use? Are these safe to use?’” Johnston said.
During DockerCon’s Wednesday keynote, Docker answered that question by announcing its GenAI Stack. GenAI Stack combines technologies and tools from Docker and partners Neo4j, LangChain and Llama. It’s a pre-configured solution with Llama providing the large language model, Neo4j providing the vector and graph databases, and the LangChain’s framework.
“We defined a GenAI stack that addresses the most common use cases in generative AI, packaged each of these technologies up as a Docker container and then put Docker Compose around it,” he said. “They don’t have to go off and search for the bits and bytes. They don’t have to cobble it together. They don’t have to configure it. It’s just ready to go.”
Developers can get going within five minutes, see the impact of their work and iterate from there, he added.
The second offering launched Wednesday was Docker AI, a generative bot designed to assist with using Docker. It’s trained on Docker’s Compose files and the library of errors that comes from Docker’s product and tools, Johnston said. There is no customer data involved, he added.
While Copilot and similar code completion AI address the 10 or 15% of an application, apps also usually leverage a database, a web server, a frontend and base images.
That remaining percentage is where Docker already plays by providing container images and Docker Compose files to define other parts of the application around the source code, he said. Docker AI, then, compliments code completion AIs by addressing that final 85-90%.
“Docker is already providing container images, Docker files and Compose files to define these other parts of the application around the source code,” he said. “What that does is gives us this wonderful opportunity to complement the Copilots and the Ghostwriters with an AI that addresses this 85% that we’re talking about.”
The AI is designed to integrate into the developer workflow, helping developers solve problems or bugs that might otherwise send them out of flow, he added. It integrates with Visual Studio Code and JetBrain’s internet development environments.
“You can think of this — not as a replacement for a human — but a mech suit for the human,” he said. “We want to keep them in the flow state, keep them iterating, building and evolving without having them jump out to some third party.”
The AI assistant is currently available through application to a Docker early access program.
