Intelligent Remediation

Actions and automations created with generative AI

Instana uses generative AI models from watsonx.ai to create actions and automations that are linked to various events you encounter.

For manual actions, use the RAG (Retrieval Augmented Generation) pattern to start watsonx.ai models. Then, create the "next best action" description as seen in the following graph:

After the description is created, use the code generation models from watsonx.ai to create automations. For example, scripts and ansible playbooks as seen in the following graph:

After you enable the feature flag for FM (foundation models) Automation, these actions and automations are going to be available in your catalog. All of them include the following watsonx tag:

Every event that is triggered by Instana undergoes analysis by the automation matching engine. This engine employs diverse algorithms to suggest actions from Instana's catalog for debugging, resolving, or debugging and resolving the event.

When you load the watsonx-powered actions, they are going to be highly recommended for events that are closely related to a specific action. The following screenshot is an example Kubernetes Pod condition status event:

To create a policy with an action to the event, click the plus icon in the Recommended Actions section. This function allows you to:

• See the policy that you created by using the action in the Automation Policies tab for any subsequent instance of this event type.
• To run this action.
• To provide feedback to the AI models that indicates that this recommendation is now accepted.

Similarly to other actions in the catalog, you must enable one or more script or ansible sensors to start the actions that are carried out in the corresponding Agent.

For scripts that include kubectl commands, make sure that the kubectl CLI is installed and configured on the target host. For more information, see Managing automation.