Introducing JFrog Fly: The World’s First Agentic Artifact Repository

AI has created a paradigm shift in software development. AI-native development teams – from small startups to enterprises like Goldman Sachs and Google – are adopting agentic development tools like Cursor and Copilot to increase the speed of code generation to a pace we’ve never seen before. But with all this new code comes a big challenge: how do you manage all these potential new releases and get the right ones deployed? 

Today’s developer stack has many AI tools, but the agentic experience falls short when it comes to creating a release or simply managing your artifacts. For AI-native teams, which are often small teams with limited resources, having to rely on legacy approaches and manual processes creates an outsized burden which creates a drag on release velocity.

As tempting as it is to “unleash” AI into your release pipelines, you need to make sure that you’re maintaining control and trust over the release process. Instead, you need a system of record for artifacts that provides agents with the context required for creating and managing releases. Only then can you guide these releases through your trusted automation with developer oversight and approval.

This is why we’ve created JFrog Fly – to make agentic releases a reality.

JFrog Fly: The Agentic Artifact Repository for Small Teams

JFrog Fly is an agentic artifact repository designed to support AI-native software development and agentic delivery workflows. It’s a zero-configuration solution that connects your binary repository to your source control and lets you manage artifacts and releases in a fully AI-native way.

How does JFrog Fly enable agentic releases? 

With JFrog Fly, every package and potential release is automatically collected and managed in one centralized place, along with all the context surrounding it. Comprehensive metadata such as commits, pull requests, change summaries, associated issues, and more is kept alongside the artifact and structured to be provided to agents via the Model Context Protocol (MCP).

Instead of working piecemeal across tools or spreadsheets to ensure you’re deploying the right thing, with JFrog Fly any release becomes meaningful by itself, can be located semantically, and operated against agentically – all from the developer IDE. Just chat “deploy the version that fixes the checkout page bug” and the agent takes care of the rest.

Wow that’s cool, what are the other capabilities and benefits of JFrog Fly?  

JFrog Fly is designed to help lean teams streamline the development and release process:

• Simple and Intuitive Experience: Fly treats artifacts as a natural extension of the coding process. For every GitHub code repository, there’s a corresponding repository in Fly that contains the packages and releases generated from the source code, and everything is bi-directionally linked.
• One Place for All Your Artifacts: Fly handles all the artifacts your team, AI assistants, customers, and runtime environments rely on, including packages, containers, and models. It’s your smart private registry.
• Distribute and Deploy with Ease: JFrog Fly provides a single place to share your releases with teammates, customers, partners, other agents, and your runtime environment. It also simplifies token management and deployment by auto-generating the necessary Helm charts, manifests, or Kubernetes descriptor files.
• Extend the AI-Native Dev Stack: Developers never have to leave their IDE. JFrog Fly delivers all the context and insights about packages and releases directly into the IDE’s AI chat, and you can use natural language commands to take action against any asset. For example, you can ask, “What changed over last week’s releases?” or “Deploy the latest build from John.”
• Zero Configuration: Fly works out-of-the-box with GitHub, AI-native coding environments, Kubernetes, and deployment tools like Argo CD. You don’t have to deal with cumbersome file editing; Fly’s config automation handles all the actions for you, including resolving and caching dependencies.
• Semantic, Agentic Releases Made Real: You no longer need to sift through version numbers to find the right fix. With all the context in one place, you can simply describe what you want to release and let your agent do the lifting.
• Accelerated Builds and Releases: Fly helps you move fast in a trusted way, both in identifying and releasing the right assets and during the build process. It organizes and manages and caches reusable packages for faster builds and easier development.

JFrog Fly as an Extension of the JFrog Platform

While JFrog Fly is a new offering, it’s built on the trusted services provided by the JFrog Software Supply Chain Platform – including Artifactory.

The JFrog Platform, with Artifactory at its core, is the trusted solution for enterprises to manage their software supply chains. It is a single system of record for secure, automated software releases that serves thousands of organizations, including most of the Fortune 100 companies.

JFrog Fly is an extension of the JFrog Platform, designed specifically for the needs of small, AI-native development teams. It introduces an AI-native user experience and removes complex enterprise-level configurations that might slow down smaller teams. JFrog Fly delivers DevOps best practices, honed from over 15 years of experience, without requiring a dedicated DevOps team. The user experience of JFrog Fly is planned to be gradually introduced into the JFrog Platform.

Ready to Fly?

Be one of the first to experience the future of AI-native software delivery – JFrog Fly is now available in Beta.

Head over to jfrog.com/fly to request access and provide us with your feedback at  fly-support(at)jfrog.com.