09.04.2018

Get started with Agent Intelligence (AI) in Incident Management.

Kim Hougaard Simonsen

Kim Hougaard Simonsen

Solution Architect

Examples of where - and how - to utilize ServiceNows AI features.

Implement or improve on your Incident Management process leveraging the latest NOW Platform capabilities based on predictive modelling and supervised machine learning, industry leading ITIL processes and no-code automation engine.

Benefits

  • Decrease response time and resolution time
  • Reduce miscategorizations and reassignments
  • Increase customer satisfaction



1 | Categorize incidents automatically based on short description using Agent Intelligence (New in Kingston)

Allow ServiceNow to fully automate categorization of incoming incidents using machine learning by teaching ServiceNow to recognize patterns in your data by analyzing a portion of your organizations historic incidents.

Auto categorizing with Agent Intelligence (AI)

Figure 1. Now Agent Intelligence is auto-categorizing your incidents for you - one less step to do.



  1. First activate Agent Intelligence on your sub-production instance.
  2. Then get your Training Set from your historic incident records in your production instance - preferably ≥30.000.
  3. Then review and specify a Predictive Model from the default Solution Definitions, in this case the Incident Categorization Solution Definition, which predicts incident category (output field) based on the short description field (input field).
  4. Apply predictions by enabling the Default Task Based Prediction business rule to allow ServiceNow to make a field value prediction before insert of the incident based on what ServiceNow learned from the Training Set.
  5. Update predictions by enabling the Update Prediction Results business rule to allow ServiceNow to learn what incident category was set by the time it was closed.
  6. Now train your Solution Definition using your Training Set. Validate Filter and Training Frequency.
  7. Then test your predictions simply by creating a couple of incident records. Are the incidents categorized as expected?
  8. Now evaluate prediction accuracy and coverage by looking at your prediction reports.
  9. Lastly retrieve and commit Update Set to your production instance.


Now your incoming incidents are categorized consistently and efficiently by ServiceNow, reducing response times and miscategorizations while improving user satisfaction.

2 | Prioritize incidents automatically based on short description using Agent Intelligence (New in Kingston)

Same goes for the Incident Prioritization Solution Definition, which predicts incident priority (output field) based on the short description field (input field).

Prioritize your incidents with Agent Intelligence (AI)

Figure 2. After this, priority is now also done by Agent Intelligence automatically.



  1. Train your Solution Definition using your Training Set. Validate Filter and Training Frequency.
  2. Then test your predictions simply by creating a couple of incident records. Are the incidents prioritized as expected?
  3. Now evaluate prediction accuracy and coverage by looking at your prediction reports from its Dashboards.
  4. Lastly retrieve and commit Update Set to your production instance.

Now your incoming incidents are prioritized consistently and efficiently by ServiceNow, reducing response- and resolution times and improving user satisfaction.



3 | Assign incidents automatically based on training set using Agent Intelligence (New in Kingston)

Automatically assign incidents using Agent Intelligence (AI)

Figure 3. From now on the incident is delegated automatically to the appropriate Assignment group as well.



  1. First review and specify the Task Assignment Solution Definition, which predicts task assignment (output field) based on the short description field (input field).
  2. Now train your Solution Definition using your Training Set. Validate Filter and Training Frequency.
  3. Then test your predictions simply by creating a couple of incident records. Are the incidents assigned to the expected assignment groups?
  4. Now evaluate prediction accuracy and coverage by looking at your prediction reports from its Dashboards.
  5. Lastly retrieve and commit Update Set to your production instance.

Now your incoming incidents are assigned consistently and efficiently by ServiceNow, reducing the time from submittal to response and resolution and thereby improving customer satisfaction.

 

 Want a dialogue with the author of this article?

Get in touch with Kim: