Your Auto-Resolution intents aren't accessible for topic mapping. Check that the prediction server i

kaushalmore
Tera Contributor

kaushalmore_0-1748507907080.png

Having an issue while configuring IAR Hr

Alert level: critical.

Your Auto-Resolution intents aren't accessible for topic mapping. Check that the prediction server is set up.
 
How can we tackle this
1 REPLY 1

Abbas_5
Tera Sage
Tera Sage

The message "Your Auto-Resolution intents aren't accessible for topic mapping. Check that the prediction server is set up in ServiceNow" indicates that the virtual agent's machine learning models, which are used for predicting topics and resolving issues, are not configured or accessibleThis is typically due to the prediction server, which houses these models, not being properly set up or active. 
 
Here's a breakdown of what to check and how to address this issue:
 
1. Verify Prediction Server Configuration:
  • Check ServiceNow documentation:
    Refer to ServiceNow's documentation on setting up and configuring the prediction server for virtual agent. This documentation will outline the necessary steps for deployment and verification.
  • Review Virtual Agent settings:
    Navigate to "All > Conversational Interfaces > Settings > Virtual Agent" and examine the ITSM Virtual Agent (IAR) settings. Look for any errors or warnings related to the prediction server's status.
  • Check the "Train" button:
    If the "Train" button is inactive, it could mean the prediction server isn't active, or there's an issue with the configuration for the specific table you are using for issue auto-resolution. Ensure the model is trained and active. 
     
2. Ensure Data and Configuration:
  • Check the number of records:
    The query used to train the model needs a minimum number of records. You can find this minimum number in the glide.platform_ml.api.min_agent_zero_records system property. 
     
  • Review your auto-resolution configuration:
    Ensure that the conditions specified in the auto-resolution configuration records return less than 300,000 records, as training the model can fail if the number of records is too high. 
     
  • Check the training status:
    Verify that the "Training Status" field displays "Solution Complete". 
     
  • Verify the "Active" property:
    If the "Training Status" is complete, make sure the "Active" property is enabled. 
     
  • Review the intent configuration:
    Ensure that the intents you are using for topic mapping are properly configured and associated with the relevant records. 
     
3. Troubleshooting:
  • Check the open_nlu_predict_log:
    This log can provide insights into why a particular intent is not being predicted for a given utterance. 
     
  • Run simulations:
    Use the "Configure and run an Issue Auto Resolution simulation" to test the matching intents based on the data from your organization. 
     
     
In essence, the key is to ensure that the prediction server is properly configured, the machine learning model is trained and active, and the intents are correctly mapped to the relevant records for issue resolution. 
 
 
If this is helpful, please hit the thumbs up button and accept the correct solution by referring to this solution in future it will be helpful to them.
 
Thanks & Regards,
Abbas Shaik