AI Catalog in Self-Managed Systems

Follow these 5 steps to go from zero to using AI Catalog in your self-managed environment.

Step	Description
	Understand the Components JFrog ML Bridge and the JFrog AI Catalog.
	Verify Prerequisites Verify your environment meets these minimal requirements before proceeding.
	Enable AI/ML Service Follow the instructions in AI/ML Service Activation to enable JFrog AI/ML.
	Install JFrog ML Bridge Enables communication between your self-managed JFrog Platform Deployment (JPD) and the JFrog AI/ML control plane.
	Verify Bridge Connection Verify JFrog ML Bridge is connected and configured.

Step 1: Understand the Components

Before beginning the installation, understand the core components of the JFrog AI/ML ecosystem and how they function in a self-managed environment.

• AI Catalog: The JFrog AI Catalog serves as your organization’s central hub for AI asset governance and visibility. It allows you to manage compliance, mitigate risks, and track AI assets. It includes:

  • Models Registry: A centralized registry to browse and manage verified AI models.
  • MCP Registry: A centralized registry to browse and manage Model Context Protocol (MCP) servers, including tool policies.
  • Skills Registry: A centralized registry to browse, manage and secure Agent Skills.
  • Shadow AI Detection: Identify and manage unapproved or "rogue" AI models within your JFrog Platform and organization.

• JFrog ML Bridge: The JFrog ML Bridge is a secure connector that runs inside your infrastructure to create a bidirectional tunnel between your self-managed JFrog Platform and cloud-hosted JFrog ML control plane.

  To enable the JFrog AI/ML features, you must install the JFrog ML Bridge. This is a specialized flavor of the standard JFrog Bridge that requires its own separate installation and includes unique modifications specifically designed to power the AI Catalog.

  Why the ML Bridge is Essential:
  In a self-managed deployment, your JFrog Platform cannot directly communicate with cloud-hosted AI/ML services. The ML Bridge solves this by:

  • Establishing Secure Connectivity: Routes all AI/ML traffic through an authenticated, encrypted, bidirectional tunnel.
  • Handling Authentication: Securely manages the credential exchange between your local JFrog Platform and the cloud-hosted JFrog ML backend.
  • Enabling AI Catalog Integration: Synchronizes model metadata and governance controls, allowing your teams to safely browse and track models while keeping your infrastructure secure.

Step 2: Verify Prerequisites

Verify your environment meets these requirements before proceeding. If any of the prerequisites is not met, setup will fail later. This prevents wasted time troubleshooting configuration issues:

1. Verify your JFrog Platform installation type, as AI Catalog supports:

   • A Kubernetes-based installation.
   • A VM installation with a containerized environment such as Docker Compose.

2. Verify JFrog Platform component versions:

   • Artifactory: 7.117.5 or higher.
   • Xray: 3.137.30 or higher.
   • Package Catalog: 1.34.6 or higher.

3. Verify license and entitlements:

   • AI Catalog and Curation entitlements.
   • Do not enable AI Catalog on Edge nodes.

4. Confirm network access. Your Artifactory must have direct outbound HTTPS access to:

   • grpc.qwak.ai or bridge.ml.jfrog.io for the JFrog ML control plane.

5. Verify JFrog admin access. You must have Platform Administrator privileges.

Step 3: Enable AI/ML Service

When implementing this procedure, you are enabling the local JFrog AI/ML frontend service on your self-managed JFrog Platform and establishing its direct connection to JFrog's cloud-hosted backend.

To enable the AI/ML service:

Follow the instructions in AI/ML Service Activation for Self-Managed Environments to enable JFrog AI/ML.

What happens:

The AI/ML service becomes visible in your JFrog Platform. Bridge functionality and configuration UI become available. You're now ready to install and configure the JFrog ML Bridge.

Step 4: Install JFrog ML Bridge

The JFrog ML Bridge is the mandatory connectivity layer that enables communication between your self-managed JPD and the JFrog AI/ML backend. The following steps explain how to install the JFrog ML Bridge. This should be done after you have activated the JFrog AI/ML service.

To install the JFrog ML Bridge:

1. Navigate to AI/ML in the JFrog Platform UI and follow the installation steps on-screen.

2. In the Internal Artifactory URL field, enter the URL that will allow the JFrog ML Bridge client (running in your Kubernetes cluster) to communicate with your JPD.

   📘

   Note

   The JFrog Bridge should be installed where it has a network connection to Artifactory.

   

3. In the Administration module, select Security > General and scroll down to Connection details.

4. Copy the Join Key and download the values.yaml file.

   

5. Copy the generated Helm command snippet and run it in the Kubernetes cluster where your JPD is installed.

   The following is an example complete Helm command snippet.

   Shell

   helm repo add jfrog https://charts.jfrog.io/ && \\
   helm repo update && \\
   helm install jfrog-bridge jfrog/bridge --set jfrogUrl="<YOUR_JPD_URL>" --set tunnel.client.bridges.jfrogml.local.url="<YOUR_INTERNAL_ARTIFACTORY_URL>" --values values.yaml --set joinKey=<JOIN_KEY>
   

   Where:

   • The jfrogUrl parameter in the Helm command snippet will update automatically according to your base URL.
   • The tunnel.client.bridges.jfrogml.local.url parameter will update automatically according to the Internal Artifactory URL that you input.
   • The <JOIN_KEY>: The join key you copied from Connection details.

6. Select the confirmation checkbox I've successfully installed the JFrog Bridge.

7. Click Next.

What happens:

The JFrog ML Bridge is installed in your Kubernetes cluster and establishes secure communication with your JPD and the JFrog AI/ML backend. The Bridge acts as a secure tunnel between your self-managed JFrog Platform and the cloud-hosted AI/ML services.

Verification: After installation, the JFrog ML Bridge status should be visible.

Step 5: Verify Bridge Connection

What to do: After the JFrog ML Bridge is installed, verify it's connected and configured.

To configure the JFrog ML Bridge and verify connection:

1. In Artifactory, go to Administration > Manage Bridges.
2. Check the JFrog ML Bridge status. The status should show as connected.

What happens:

Bridge is verified as installed and connected to your JPD. The secure tunnel to JFrog's AI/ML backend is established and ready for AI Catalog initialization.

If Bridge is not connected:

• Check network connectivity to Artifactory from the Bridge pod
• Verify outbound HTTPS access to grpc.qwak.ai or bridge.ml.jfrog.io
• Review the JFrog ML Bridge pod logs for connection errors
• Ensure the Join Key and other parameters were correct during installation

You are now ready to start using AI Catalog. For more information, see Getting Started with the AI Catalog.

Frequently Asked Questions

This section provides answers to frequently asked questions.

Related Topics

• JFrog AI Catalog Overview
• AI/ML Service Activation for Self-Managed Environments
• Shadow AI Detection and Governance
• Troubleshoot AI Catalog