Creating IBM watsonx.ai integrations

• For this integration, you must set up an IBM watsonx.ai project running on IBM Cloud (SaaS) in the Dallas or Frankfurt regions. On-prem installations of watsonx.ai are not supported for this integration. The project must be for use only by IBM Cloud Pak for AIOps under the Developer access panel.

  Tip:

  • To set up watsonx.ai for the integration, see the following topics in the watsonx.ai documentation:

    • Setting up the IBM watsonx platform
    • Setting up the watsonx Runtime services.
      For higher rate limits that might be required for the Cloud Pak for AIOps integration, you are suggested to choose at least the watsonx.ai Essentials plan.
    • Associating watsonx.ai runtime with watsonx.ai in IBM Cloud
    • Administering watsonx.ai projects

  • After you set up the watsonx.ai instance, verify that the model inference is set up correctly by trying a simple prompt, for example, "Hello! How are you?" in the watsonx.ai Prompt Lab. Make sure that you use the Granite 4-h-small model. No additional steps are necessary.

• From the watsonx.ai project that you set up, collect the following details:

  • Project ID: To find the project ID, see Finding the project ID in the watsonx.ai documentation.
  • watsonx.ai URL: To find the URL for watsonx.ai, go to the Developer Access panel on the watsonx.ai home page.
  • IBM Cloud API token: To find the IBM Cloud API token, see Managing user API keys in the IBM Cloud documentation. The IBM Cloud API token should be used only by the IBM watsonx.ai integration in IBM Cloud Pak for AIOps.

For more information about working with IBM watsonx.ai integrations, see the following sections:

• Enabling the feature
• Disabling the feature
• Creating IBM watsonx.ai integrations
• Editing IBM watsonx.ai integrations
• Deleting IBM watsonx.ai integrations

To enable the configuration UIs to use the generative AI feature, you must create a feature flag configmap:

1. Install the latest IBM Cloud Pak for AIOps.

2. In Cloud Pak for AIOps namespace in the Red Hat OpenShift UI, go to Home > Search.

3. Under the Resources dropdown, search for ConfigMap.

4. Click the Create ConfigMap button.

5. Paste the following YAML and create the configmap:

   apiVersion: v1
   kind: ConfigMap
   metadata:
     name: feature-flag-configmap
   data:
     AIOPS_GENERATIVE_AI: enabled
   

6. After a few seconds you should see that the AI Models UI and Connections UI pods have restarted.

To disable the feature:

1. Delete the Incident summarization AI Model in the AI models UI.
2. Delete the watsonx.ai integration.
3. Disable the feature flag by setting the key to disabled.

To create an IBM watsonx.ai integration, complete the following steps:

1. Log in to IBM Cloud Pak for AIOps console.

2. Expand the navigation menu (four horizontal bars), then click Define > Integrations .

3. On the Integrations page, click Add integration.

4. Navigate to the Generative AI category, find and click the IBM watsonx.ai tile.

5. On the side-panel, review the prerequisites.

   

6. Click Get Started.

7. On the Add integration page, define the general integration details:

   • Name: Type in the display name of your integration.

   • Description: Provide an optional description for the integration.

     

   • watsonx.ai Project ID: Enter the ID of the watsonx.ai project that you want to use. See Finding the project ID in the watsonx.ai documentation.

   • watsonx.ai Runtime plan: Select from the following options:

     • Lite
     • Essentials or Standard
     • Custom

   • watsonx.ai URL: Enter the URL for watsonx.ai from the Developer Access panel on the watsonx.ai home page.

   • IBM Cloud API key: Enter the API key created in IBM Cloud from an account that has access to watsonx.ai and the watsonx.ai project. See Managing user API keys in the IBM Cloud documentation.

   • IBM Cloud IAM URL: Enter the IBM Cloud default access management URL for your account. This field usually does not need to be updated. .

8. Click Next.

9. On the Configure Proxy (Optional) page complete the following fields:

   • Connect to IBM watsonx.ai via a proxy server: Set the toggle button to on.

   • URL of HTTP proxy server: Enter a URL for the HTTP proxy server.

   • Port number of proxy server: Enter a port. The default is 8080.

   • Connect to proxy server with username and password: Set the toggle button to on.

   • Proxy username: Enter username that you want to log on to the proxy server with.

   • Proxy password: Enter password associate with the username that you want to log on to the proxy server with.

     

10. Click Next.

11. On the Configure token usage metering Optional page complate the following fields:

    • Limit IBM watsonx.ai token usage: Toggle limiting maximum watsonx.ai token usage for the integration.

    • Maximum IBM watsonx.ai token usage: Sets a cap on daily token usage for this integration. Limits reset daily at UTC midnight and there is no limit if not enabled. Daily token usage may slightly surpass the maximum limit since requests stop processing only when the limit is reached.

12. Click Done.

    You will see the new integration appear in the list of IBM watsonx.ai integrations.

After you create your integration, you can edit it. To edit a integration, complete the following steps:

1. Log in to IBM Cloud Pak for AIOps console.

2. Expand the navigation menu (four horizontal bars), then click Define > Integrations .

3. Click the IBM watsonx.ai integration type on the Manage integrations tab of the Integrations page.

4. On the IBM watsonx.ai integrations page, click the name of the integration that you want to edit. Alternatively, you can click the options menu (three vertical dots) for the integration and click Edit. The integration configuration opens.

5. Edit your integration as required. Click Save when you are done editing.

If you no longer need your IBM watsonx.ai integration and want to not only disable it, but delete it entirely, you can delete the integration from the console.

To delete a integration, complete the following steps:

1. Log in to IBM Cloud Pak for AIOps console.

2. Expand the navigation menu (four horizontal bars), then click Define > Integrations .

3. Click the IBM watsonx.ai integration type on the Manage integrations tab of the Integrations page.

4. On the IBM watsonx.ai integrations page, click the options menu (three vertical dots) for the integration that you want to delete and click Delete.

5. Enter the name of the integration to confirm that you want to delete your integration. Then, click Delete.

Your integration is deleted.