---
sourceDocument: Xanadu Enable AI
sourceDocumentLink: https://servicenow-prod.fluidtopics.net/r/xanadu/intelligent-experiences

 Release :

    - xanadu

ft:locale :

    - en-US

ft:publication_title :

    - Xanadu Enable AI

ft:clusterId :

    - platai

bundleId :

    - platai

workflow :

    - Platform


---

# Use Predictive Intelligence in Workflow Studio with ML actions

# Use Predictive Intelligence in Workflow Studio with ML actions {#ariaid-title1}

* Release version: Xanadu
* 
* Updated May 12, 2025
* 
* ![](https://www.servicenow.com/docs/portal-asset/ico-clock) 6 minutes to read

Use Predictive Intelligence actions in Workflow Studio to create flows that incorporate your model predictions.

## Before you begin

* Make sure the following plugins are activated in your instance: Predictive Intelligence (com.glide.platform_ml) and Predictive Intelligence for Workflow Studio (com.snc.ml_flowdesigner).
* Create or use an existing trained Predictive Intelligence solution.
* Roles required: ml_admin or admin, and flow_designer or delegated_developer.
{#use-predictive-intelligence-flow-designer-ml-actions__ul_sh5_z44_mnb}

## About this task

Workflow Studio enables you to automate complex processes. The first thing to identify is what process you want to automate. In this example scenario, you're automating the assignment of a Category to an
incident record. When you complete the flow, the next incident record created in your instance updates the Category field in the record based on the text entered in the Short description field.

You can deploy any active and trained classification, similarity, or regression ML solution in your flow, as appropriate for your use case.

In this example procedure, you create a flow that implements the ml_incident_categorization solution in a Workflow Studio action. You can find this solution by searching on the ML Solutions \[ml_solution\] table, as shown in the image below. Confirm that the solution you use has been trained and its
Active value is set to true.

For more information on how to use Flow Designer in Workflow Studio, see . For information about the Actions included in Predictive Intelligence for Workflow Studio, see the Spoke actions table in .

## Procedure

1. Navigate to AllProcess AutomationWorkflow Studio.
2. Select NewFlow.  
3. On the Let's get the details for your flow screen, configure the following fields.  
   Expand Show additional properties to view all fields.{#use-predictive-intelligence-flow-designer-ml-actions__table_lrd_1xd_f2c__entry__2}

   | Field | Description |
   |-|-|
   | Flow Name | Provide a name for the flow. In this scenario, you enter <kbd class="ph userinput">Auto-assign Category to Incident</kbd>. |
   | Description | Enter a brief summary description of what the flow delivers. For example, you enter the following: <kbd class="ph userinput">When an incident is created, it automatically triggers this flow, which uses ML Solutions to predict the correct Category for the incident.</kbd> |
   | Application | Select Global. |
   | Protection | Select --None-- or Read-only. In this scenario, you select --None--. |
   | Run As | Select User who initiates session. |
   | Run with role(s) | Leave blank. |
   | Flow priority default | Medium (default). |
   [ ]

   {#use-predictive-intelligence-flow-designer-ml-actions__table_lrd_1xd_f2c}
4. Select Build flow.  
   The Flow screen appears, showing the Auto-assign Category to Incident name you assigned to the flow. If a Getting started screen appears, select Skip tour.
5. In the TRIGGER section of the Flow screen, configure the following fields to create a trigger for the flow.
   1. Select a Trigger: Select Record, then from the list of possible options for Record, select Created.  
   2. Table: Select Incident \[incident\].  
      Note:  
      After you configure both the Trigger and Table fields, record data pills appear in the Data section of the screen so you can use them in your flow.  
   3. Condition: Select Add filters if you want to add any conditions to the flow.
   4. **Optional:** Open the Advanced Options panel to view additional conditions you can apply to the flow.
   5. **Optional:** To close the panel, select Advanced Options.
   6. Select Done.
   {#use-predictive-intelligence-flow-designer-ml-actions__substeps_cl5_c4q_d4b}
6. In the ACTIONS section of the screen, configure the following fields to create a Classification Prediction action.
   1. Action tab: Select ActionPredictive IntelligenceClassification Prediction.  
      Select the information icon (![Information icon.]()) to see the description of a Classification Prediction.
   2. Solution \[ML Solution\]: Select ml_incident_categorization.
   3. Top N: Enter <kbd class="ph userinput">3</kbd> for the example scenario.  
      When you enter a number, such as 3, the system uses the top three ML predictions that have the highest prediction confidence score. If you don't enter anything, the system sets the default value to 1.
   4. Input Record: Drag and drop your Trigger → Incident Record data pill into the Input Record field.  
      The Action, Solution Name, Top N, and Input Record values provide a base for the Category prediction.  
      Note:  
      The data pill you drop into this record must also be a record. For example, don't try to drop a table pill or a date/time pill into the Input Record field.
   5. Select Done.  
      Result: The Classification Prediction action is completed in the flow and its data pills appear in the Data section of the screen.  
   {#use-predictive-intelligence-flow-designer-ml-actions__substeps_bbf_s5k_qnb}
7. In the ACTIONS section of the screen, use the following steps to create actions and flow logic for the incident's Prediction Results.  
   Note:  
   Although you can use a loop to iterate through every prediction result, the scenario shown in this documentation uses a relatively small number of actions. For more advanced flow configurations, see the [Flow
   Designer](https://www.servicenow.com/docs/access?context=flow-designer&version=xanadu&pubname=xanadu-application-development&ft:locale=en-US).
   1. For each item in list of items: Drag and drop the Prediction Results data pill into the Items field.  
      Note:  
      In order to access the list of items in the Regression Prediction action, you don't need the For Each Item in flow logic.
   2. Select Done.  
      Result: The Prediction Results action is started in the flow and its data pills appear in the Data section of the screen.
   {#use-predictive-intelligence-flow-designer-ml-actions__substeps_nbm_jvq_d4b}
8. In the ACTIONS section of the screen, select ActionPredictive IntelligencePI Confidence Check.  
   The PI Confidence Check is a tool you can use to compare values in a flow. In this use case, it compares prediction result values, and the output from the check is either True or False.  
9. Drag and drop the confidence data pill into the Predicted Number from Predictive Intelligence field.
10. Enter <kbd class="ph userinput">50</kbd> in the Comparison Threshold field.  
    In this example scenario, you enter the number 50, which tells the system to use predictions that have a confidence score above 50%.
11. Select Done.  
12. Select Flow LogicIf to add a condition to the flow.  
13. Configure the following fields to define the first part of the condition flow logic.  
    * Condition: Enter a name for the condition that defines what it does. In this example scenario, you enter <kbd class="ph userinput">Confidence greater than 50</kbd>.
    * Condition 1: Drag and drop the Confidence To Predict data pill into the field. Select is, and enter the value <kbd class="ph userinput">True</kbd>. This step completes the first part (the antecedent) of the condition flow logic.
    * Select Done.

    {#use-predictive-intelligence-flow-designer-ml-actions__ul_rpg_nrk_l4b}  
14. Select Action and enter <kbd class="ph userinput">worknote</kbd> into the search field.
15. Select ITSMAdd Worknote to add a work note as the second part (the conclusion) of the condition.  
16. Configure the following fields to define the second and final part of the condition flow logic.  
    * Action: As a result of Step 14 above, Add Worknote appears automatically in this field.
    * task \[task\]: Drag and drop the Incident Record data pill into the field.
    * work note: Drag and drop the predicted_value data pill into the field. This step completes the condition flow logic conclusion.
    * Select Done.

    {#use-predictive-intelligence-flow-designer-ml-actions__ul_tqh_rvz_l4b}  
17. Select Action and enter <kbd class="ph userinput">update record</kbd> into the search field.
18. Select Update Record.  
19. Configure the following fields to update the Incident Record.  
    * Action: As a result of Step 16 above, Update Record appears automatically in this field.
    * Record: Drag and drop the Incident Record data pill into this field.
    * Table: Select Incident \[incident\].
    * Fields: Select Category. Then drag and drop the predicted_value data pill into this field, next to the Category value.
    * Select Done.  
    {#use-predictive-intelligence-flow-designer-ml-actions__ul_ocn_zyz_l4b}
20. Select Save.
21. Select Activate.  
    Result
    * Your Auto-assign Category to Incident flow is activated and complete.  
    * It also appears as published in the Flows column on the Workflow Studio home screen.  
    {#use-predictive-intelligence-flow-designer-ml-actions__ul_j41_5v1_m4b}
22. Navigate to Incidents.
23. Select New to create a test incident record in the Incidents table.  
    In this example scenario, you create record INC0010011.  
24. In the record you created, note in the following image that the Category value is set to Inquiry / Help.
25. In the Short description field, enter <kbd class="ph userinput">Email not working</kbd>.
26. Select Submit.  
    Result  
    The system updates the incident record to show that its Category value has changed from Inquiry / Help to Email.
    * 
    * 
    {#use-predictive-intelligence-flow-designer-ml-actions__ul_udm_wt1_m4b}
**Related topics**   

* 

*[\>]: and then