Configuring a Coleman AI step in workflow

  1. In the workflow diagram, add a Coleman AI step at the point where a task must be executed.
  2. Select the Coleman AI element from the workflow diagram.
  3. Specify this information in the properties pane:
    Name
    Specify a name for this workflow step.
    Description
    Specify a description for this workflow step.
    Coleman Task

    Select the task type to start:

    • Reload Datasets: Trigger the reloading of datasets for the specified quest name from Data Lake.

    • Retrain Model: Trigger the retraining of the machine-learning model for the specified quest name.

    • Update Production Quest: After a quest model is retrained, use this task to request updating the production quest.

    • Redeploy Endpoint: After a production quest is updated, use this task type to redeploy its endpoint.

    Note: For each Coleman AI machine learning task you must add a new step to the workflow. To have a complete orchestration, use all four task types for the same quest. Use them in the same order as they are described earlier in the list. With the workflow modeler you can use any number of Coleman AI steps and it does not validate the sequence.
    Quest
    Specify the name of the quest for which the selected Coleman task must be executed. Select one of these options:

    • Quest Name: select a quest name from the list provided by the Coleman AI backend. Only the quests with an existing endpoint are available.

    • Quest Name from Parameter: select the name of a workflow parameter of type String or Code. At runtime, this parameter must contain a valid quest name. This can be a constant value or a workflow input parameter.

    Service Account

    Import the service account obtained from Infor Ming.le User Management. The credentials in this service account are used to call the Coleman AI API to execute the Coleman task.

    Click IMPORT to select the service account file and load it into the workflow step. After the import, the description of the service account is displayed. You can use the same service account in several Coleman AI steps.

  4. Click the Test tab to start a validation of the configuration for the Coleman task, quest name and service account.

    If you have selected the option Quest Name from Parameter, you must first specify the name of a quest, then click Test.

  5. Click the Settings tab to specify how to handle errors from the Coleman task execution.
    1. Select the The workflow continues option to allow the workflow to continue to the next step in case of an execution error. Optionally, you can collect the error response code and error message that are returned:
      • Select the Error response code check box and select a workflow parameter of type String, Decimal or Integer.
      • Select the Error message check box and select a workflow parameter of type String.
    2. Select the The workflow fails option to allow this workflow instance to fail in case of an execution error.
  6. Click Save and activate the workflow.