Skip to tutorial content
Rate module Not yet rated

Predictive Intelligence for Categorization and Assignment

4 min read
On this page

In this lesson, you will learn:

  • What Predictive Intelligence does
  • How it learns from old incidents
  • What confidence means
  • What SNWR needs before using it

🔁 Quick look back: Last time: Impact Analysis for Changes. You learned how to use a CI’s dependency map to see what a change will affect.

1 “Every agent picks a different category”

SNWR’s agents pick categories in different ways.

  • “VPN keeps disconnecting” ends up as Network, Software or Inquiry.
  • Wrong categories send incidents to the wrong team.
  • Assignment rules can’t catch every wording.

🎯 The problem: People and fixed rules miss many ways of saying the same thing.

👉 The fix: let a model learn from past incidents, and suggest the category and group.

2 How it works

Key term: Predictive Intelligence

ServiceNow’s machine learning feature that learns from past records to make suggestions.

Key term: Model

What the system learns from past records, used to make guesses about new ones.

Key term: Confidence

How sure the model is. SNWR only uses suggestions above a set level.

Predictive Intelligence on Sam's incident. 1 Sam writes: VPN keeps disconnecting. 2 The model, trained on thousands of past SNWR incidents, reads the text. 3 It suggests category Network, subcategory VPN, confidence 91%, and group SNWR Network Support, confidence 88%. 4 Above SNWR's threshold, so the fields are filled. Below it, the agent decides.
Figure 1. The model reads the words and suggests where it goes.

3 Before SNWR uses it

  • It needs lots of good history. SNWR has only a few weeks after go-live.
  • Old incidents must have correct categories, or it learns mistakes.
  • Plan it for a few months after go-live.

📦 Entitlement check: Predictive Intelligence and its models depend on your package and release. Check what SNWR’s ITSM Advanced includes before planning.

⭐ Remember: A model is only as good as the records it learns from.

4 Quick recap

  • Predictive Intelligence learns from past records
  • It suggests category and assignment group from the text
  • Confidence decides if a suggestion is used
  • It needs lots of correct history
  • SNWR plans it a few months after go-live

5 Check yourself

Think of your answer first. Then tap to check.

1 What does the model read to make its guess? Show answer

The short description, like “VPN keeps disconnecting”.

2 The model is 40% sure. What happens at SNWR? Show answer

It’s below the threshold, so the agent decides.

3 Why wait after go-live? Show answer

The model needs lots of good history to learn from.

4 What if old incidents have wrong categories? Show answer

The model learns the mistakes.

Nice work! You know how a model can route incidents. 🎉
Next up: Now Assist for ITSM

Was this lesson helpful?

Have a question or doubt?

Ask here and get help from instructors and fellow learners.

0 / 300

Only logged-in learners can post.