Use AI Skills in API Tasks

Optimize your automations by utilizing our cutting-edge task execution infrastructure in API Tasks, and take advantage of faster executions of AI Agents using AI Skills with the generative AI capability.

Enhance your API Tasks by leveraging the generative AI capability of AI Skills and reduce the cost and improve performance of automations while integrating with generative AI models in a scalable, safe, and governed environment.
Important: This feature is available on Cloud only, and requires the Enterprise Platform license.

See Enterprise Platform.

Take advantage of using API Tasks to avoid the additional steps of setting up Bot Runners or a device pool for executing the AI Skills. A Pro Developer or Citizen Developer would use this feature to create an API Task integrated with an AI Skill.

AI Skills package integration with API Tasks

This is an overview of tasks you would perform in the following order to integrate AI Skills in API Tasks:
  1. You would first create a Model connection in AI Agent Studio, and then use the Model connection in an AI Skill.

    See Create and manage Model connections.

  2. Next, create an AI Skill connecting to the Model connection you created.

    AI Skills are created by the Pro Developer and assigned for reuse to the Citizen Developer or Pro Developer. This can be directly used within an automation thus leveraging the generative AI capability by executing the action and making a call to the connected foundational model, via Model connection. See Create AI Skills.

  3. Next, create an API Task and use the AI Skills package > Execute action available for use.

    See Create an API Task.

  4. In the API Task choose the AI Skill you created.
  5. When you select the AI Skills package > Execute action in the API Task, you would see the Prompt Input variable that gets passed from the API Task to the AI Skill.

    A container is deployed for an API Task that invokes the AI Skills package within the container that executes the Execute action which makes a call to the linked foundational model via Model connection to retrieve a response. This response gets assigned to the output variables.

  6. Next, run the API Task.

    See Run API Tasks.

Known product behavior

In Automation 360 v.34 release, AI governance Audit logs are not supported for API Tasks. This is specific to the API Tasks that are run in real-time mode (for attended automations).

See API Task real-time endpoint.

Some important references

These are some important references: