Predictive Intelligence & ITSM in ServiceNow’s Xanadu release revolutionizes IT Service Management (ITSM) by enhancing automation and efficiency. AI-driven features can reduce incident resolution time by 30%, improve service desk efficiency by 25%, and enhance accuracy in ticket categorization by 40%. Implementing Predictive Intelligence optimizes workflows for Knowledge Management, Agent Assist, Incident Management, Service Catalog, and Change Management.

This guide is a step-by-step approach to configuring Predictive Intelligence in ServiceNow.

---

Prerequisites

1. Activate Predictive Intelligence Plugins

• Search for Predictive Intelligence and click Install.
• Ensure that the plugin is successfully activated.

2. Assign Roles

• Assign appropriate roles such as admin or ml_admin to users configuring Predictive Intelligence.

Predictive Intelligence & ITSM in ServiceNow’s Xanadu release revolutionizes IT Service Management (ITSM) by enhancing automation and efficiency. AI-driven features can reduce incident resolution time by 30%, improve service desk efficiency by 25%, and enhance accuracy in ticket categorization by 40%. Implementing Predictive Intelligence optimizes workflows for Knowledge Management, Agent Assist, Incident Management, Service Catalog, and Change Management.

This guide is a step-by-step approach to configuring Predictive Intelligence in ServiceNow.

---

Prerequisites

1. Activate Predictive Intelligence Plugins

• Search for Predictive Intelligence and click Install.
• Ensure that the plugin is successfully activated.

2. Assign Roles

• Assign appropriate roles such as admin or ml_admin to users configuring Predictive Intelligence.

---

A. Setting Up Predictive Intelligence for Knowledge Management

1. Activate the Knowledge Management Plugin

•  Predictive Intelligence for Knowledge Management plugin (com.snc.knowledge_ml)

2. View and Set-Up Predictive Intelligence Homepage

• Navigate to Predictive Intelligence > Homepage.

3. Train the Machine Learning Model

Using the Similarity Model in ServiceNow for Identifying Similar Knowledge Articles

The Similarity model in ServiceNow’s Predictive Intelligence framework allows the system to find records that are contextually similar to an incoming record. This can help agents quickly resolve incidents by suggesting relevant knowledge articles based on past cases.

Understanding the Similarity Model

The Similarity Model in ServiceNow’s Predictive Intelligence framework identifies contextually similar records, enabling agents to resolve incidents faster. By leveraging past case data, this AI-driven feature suggests relevant knowledge articles to assist in troubleshooting.

How It Works

• The model analyzes textual similarity between records.
• It builds a vocabulary of words from trained records to compare and rank matches.
• This enhances knowledge recommendations, reducing manual searches and improving resolution times.

For instance, if an agent logs an incident about VPN connectivity issues, the system can automatically suggest VPN troubleshooting articles based on previous cases.

---

Setting Up the Similarity Model for Knowledge Articles

Accessing Predictive Intelligence

Navigate to Predictive Intelligence > Solution Definitions and select New to create a Similarity Solution.

Defining the Similarity Model for Knowledge Articles

• Solution Name: ml_sn_global_articles_similar_to_incidents
• Description: Identifies knowledge articles similar to incident descriptions.
• Table to Train On: incident
• Target Table: kb_knowledge (Knowledge Articles)

Configuring Input and Output Fields

Input Fields (from the Incident table):

• short_description
• description

Target Fields (from the Knowledge Article table):

• short_description
• text

These fields ensure that incident descriptions are matched with relevant knowledge articles, improving search efficiency.

Training the Model

Click Train Model to initiate the training process. The system will extract and process textual data from existing incidents and knowledge articles. Training time depends on data volume and complexity.

Verifying Training Results

Navigate to Predictive Intelligence >Homepage, select Similarity.

ml_sn_global_articles_similar_to_incidents, and click to View Similarity Examples to check if the model is correctly matching knowledge articles to incidents.

---

Scheduled Training & Maintenance

• Next Scheduled Training: 2025-03-19
• Regular training ensures that accuracy improves as new data is added.
• You can manually trigger retraining by selecting Train Now.

---

Testing the Predictive Knowledge Model in a Live Incident Scenario

In this example, I created a knowledge article titled “How to Configure Predictive Intelligence for ITSM.” I then tested the system by creating an incident with a vague short description: “How can I improve search on Knowledge?” Predictive Intelligence successfully identified the relevant knowledge article I had just created and surfaced it under “Related Search Selection.”

This demonstrates how an IT user—whether submitting their own ticket or creating one on behalf of someone else—could access relevant knowledge before even completing the incident submission. This proactive approach can improve efficiency and reduce the need for additional follow-ups.

Working a New Incident

The ITIL user navigates to the Service Operations Workspace and locates the newly created incident. She then clicks on the “Agent Assist” icon on the right and is instantly presented with a relevant knowledge article—introducing her to an option she might not have been aware of.

Predictive Intelligence enhances this process by automatically associating Knowledge Search Options, even when the user isn’t sure what additional solutions to explore. This proactive assistance streamlines issue resolution and improves efficiency.

---

Improving Accuracy If Needed

• Retrain the model with additional data for better precision.
• Refine input fields by adding more detailed descriptions for improved matching.

---

Summary

By implementing ml_sn_global_articles_similar_to_incidents, ServiceNow automatically recommends relevant knowledge articles based on past incidents. This functionality enhances resolution speed, reduces manual searches, and improves overall IT support efficiency.

With continuous training and monitoring, Predictive Intelligence will evolve, delivering even more precise and helpful recommendations over time.

Other Resources Predictive Intelligence & ITSM

• Configuration tips for Predictive Intelligence
• Create and train a classification solution
• Create and train a similarity solution
• Explore Predictive Intelligence
• Predictive Intelligence
• Predictive Intelligence for Knowledge Management
• Predictive Intelligence Workbench integration and customization
• Requirement Builder for ITSM
• Task Intelligence for ITSM
• Testing and monitoring predictions
• Training Machine Learning solutions

---

A. Setting Up Predictive Intelligence for Knowledge Management

1. Activate the Knowledge Management Plugin

•  Predictive Intelligence for Knowledge Management plugin (com.snc.knowledge_ml)

2. View and Set-Up Predictive Intelligence Homepage

• Navigate to Predictive Intelligence > Homepage.

3. Train the Machine Learning Model

Using the Similarity Model in ServiceNow for Identifying Similar Knowledge Articles

The Similarity model in ServiceNow’s Predictive Intelligence framework allows the system to find records that are contextually similar to an incoming record. This can help agents quickly resolve incidents by suggesting relevant knowledge articles based on past cases.

Understanding the Similarity Model

The Similarity Model in ServiceNow’s Predictive Intelligence framework identifies contextually similar records, enabling agents to resolve incidents faster. By leveraging past case data, this AI-driven feature suggests relevant knowledge articles to assist in troubleshooting.

How It Works

• The model analyzes textual similarity between records.
• It builds a vocabulary of words from trained records to compare and rank matches.
• This enhances knowledge recommendations, reducing manual searches and improving resolution times.

For instance, if an agent logs an incident about VPN connectivity issues, the system can automatically suggest VPN troubleshooting articles based on previous cases.

---

Setting Up the Similarity Model for Knowledge Articles

Accessing Predictive Intelligence

Navigate to Predictive Intelligence > Solution Definitions and select New to create a Similarity Solution.

Defining the Similarity Model for Knowledge Articles

• Solution Name: ml_sn_global_articles_similar_to_incidents
• Description: Identifies knowledge articles similar to incident descriptions.
• Table to Train On: incident
• Target Table: kb_knowledge (Knowledge Articles)

Configuring Input and Output Fields

Input Fields (from the Incident table):

• short_description
• description

Target Fields (from the Knowledge Article table):

• short_description
• text

These fields ensure that incident descriptions are matched with relevant knowledge articles, improving search efficiency.

Training the Model

Click Train Model to initiate the training process. The system will extract and process textual data from existing incidents and knowledge articles. Training time depends on data volume and complexity.

Verifying Training Results

Navigate to Predictive Intelligence >Homepage, select Similarity.

ml_sn_global_articles_similar_to_incidents, and click to View Similarity Examples to check if the model is correctly matching knowledge articles to incidents.

---

Scheduled Training & Maintenance

• Next Scheduled Training: 2025-03-19
• Regular training ensures that accuracy improves as new data is added.
• You can manually trigger retraining by selecting Train Now.

---

Testing the Model in a Live Scenario

Checking Similar Knowledge Articles for an Existing Incident

Open an existing incident (e.g., VPN issue), review the Similar Knowledge Articles section, and ensure that relevant articles appear in the suggestions.

Creating a New Incident

Navigate to Incident > Create New, enter a detailed description of the issue, save the incident, and confirm whether Predictive Intelligence suggests relevant knowledge articles.

---

Improving Accuracy If Needed

• Retrain the model with additional data for better precision.
• Refine input fields by adding more detailed descriptions for improved matching.

---

Summary

By implementing ml_sn_global_articles_similar_to_incidents, ServiceNow automatically recommends relevant knowledge articles based on past incidents. This functionality enhances resolution speed, reduces manual searches, and improves overall IT support efficiency.

With continuous training and monitoring, Predictive Intelligence will evolve, delivering even more precise and helpful recommendations over time.

Other Resources Predictive Intelligence & ITSM

• Configuration tips for Predictive Intelligence
• Create and train a classification solution
• Create and train a similarity solution
• Explore Predictive Intelligence
• Predictive Intelligence
• Predictive Intelligence for Knowledge Management
• Predictive Intelligence Workbench integration and customization
• Requirement Builder for ITSM
• Task Intelligence for ITSM
• Testing and monitoring predictions
• Training Machine Learning solutions