Using Predictive Intelligence for Major Incidents

[matthather]

Hi,

I'm just starting out with Predictive Intelligence and in particular, using it to predict major incidents. Is anybody using it for this currently and what's your experience with it? 

Matt

[nschneid]

I am also working on this and have struggled to make progress. The documentation from ServiceNow is a bit misleading... it talks about predicting Major Incidents, but if you read more, it is about finding similar tickets to EXISTING major incidents (not predicting new Major Incidents). 

I currently have a workflow that uses a similarity model to identify similar tickets opened in a 24 hour rolling period. If X number of similar tickets are opened in that 24 hour period then create a parent ticket, attach the similar children tickets, and do some other things. 

My Issue: The similarity model runs in the workflow after an incident creation trigger. To run the ML model step you must have a special ML related role (otherwise you get a permissions error). Our developers do not want to give this ML role to all users, which does not make sense because in theory a wide variety of users could open up incidents related to a major incident.

I am currently stuck and brainstorming other options to predict Major Incidents. Please let me know if you have had any ideas after your initial post. 

[Eric Pleines]

I believe in the flow settings you can define "run as" and then - instead of "user who triggered" - you can set "system". This may resolve your permissions issues. When creating the parent ticket, you could still set field "created by" to the user who triggered the flow. So the only change would be that in the flow executions it says run by system and not by user.