The benefits that AI will bring to enterprises will manifest as a gradual transformation rather than a sudden change. Much like transformative technologies of the past, such as cloud computing, mobile, or the internet, organizations that can integrate these technologies at a fundamental level will see significant advantages. Similar to its predecessors, AI as a transformative technology requires infrastructure, tooling, and practices to support its adoption.
At Jozu, our goal is to empower organizations to deploy and operate AI applications in production. We have open-sourced KitOps, which includes ModelKits—an OCI-compliant package containing everything needed to integrate with a model or deploy it to production—and the Kit CLI, which manages ModelKits on any OCI-compliant repository. Since launching KitOps a few months ago, we have experienced significant interest, as evidenced by the increasing download numbers, which have surpassed a thousand weekly downloads. We have also partnered with several companies in the US and Europe to help them adopt KitOps into their AI/ML pipelines.
One recurring piece of feedback we have received is that while organizations appreciate using their existing OCI registries, they would prefer a ModelKit-first registry experience. Today, we are launching the preview of Jozu Hub, our SaaS registry with a ModelKit-first experience. Jozu Hub is designed to work with ModelKits and provide essential information about a ModelKit, such as datasets and models, at your fingertips. It allows you to see the differences between ModelKit versions and easily spot essential security information, such as provenance.
Today’s launch is just the beginning. We have more features in the works, such as private repositories, more options for search, inference container images, and many more. We have not forgotten organizations that cannot use our SaaS service. We will also release a version of Jozu Hub that you can run in your infrastructure, which will work with your existing OCI registry.
If you are interested in the ModelKit-first experience of Jozu Hub, visit our beta release at jozu.ml. If you are interested in storing your own ModelKits, sign up for early access. If you are an organization that would prefer to run your own Jozu Hub, contact us at info@jozu.com, and in the meantime, sign up for early access to test the features.
Top comments (4)
This sounds promising! How do you see Jozu Hub evolving in the next year? Also, great job on surpassing a thousand weekly downloads for KitOps!
Thank you for your kind words! We’re thrilled to see our community growing.
In the near future, we are planning to simplify the process of deploying inference microservices using ModelKits.It will be so intuitive that you’ll barely have to think about it.
Looking further ahead, it’s not hard to envision, with ModelKits and their ability to encapsulate all necessary artifacts for the AI/ML lifecycle, Jozu Hub will offer seamless interaction and control over the entire lifecycle. Best of all, this will be accomplished without forcing you to abandon your preferred tools and workflows as it is our way.
Congratulations on the release
Thank you Ayush!
Feel free to signup for our early access to stay up-to-date on what we're working on.
Jouz.com/hub