ITSM Predictive Intelligence Workbench implementation

Release version: Yokohama

Updated January 30, 2025

2 minutes to read

Summarize

Summarized using AI

Summary of ITSM Predictive Intelligence Workbench implementation

The ITSM Predictive Intelligence Workbench enables ServiceNow customers to use machine learning to enhance their IT Service Management (ITSM) processes by training and implementing predictive models. These models augment existing workflows to optimize business operations. However, starting with the Yokohama release, this Workbench is deprecated and will no longer be supported or enhanced. Customers are advised to migrate to theTask Intelligence for ITSMapplication to continue leveraging predictive intelligence capabilities.

Show full answer Show less

Key Features

Use Case Templates: Users with the piwbadmin or piwbmanager roles can explore and select from prebuilt use case templates to create predictive models. Templates come in two types:
- Guided Setup: Provides a comprehensive, step-by-step implementation process for easier adoption.
- Classic Setup: Offers a more manual configuration experience with links to relevant documentation and the Predictive Intelligence application.
Pretrained Models: Some templates include pretrained models based on your data, allowing you to skip directly to evaluation and integration phases if the model’s predicted accuracy meets your needs. You can tune or create new models if needed, and update the use case’s name and description later.

Model Creation and Implementation Process

The Workbench guides customers through several phases to create and deploy predictive models:

Create and Train Models: Define parameters to generate models trained on your specific data. Multiple models can be created and refined to achieve an optimal balance of coverage and precision.
Test Models: Evaluate prediction accuracy using single or batch testing to identify the best-performing model.
Integrate the Best Model: Deploy the selected model into production workflows to augment your ITSM processes.

For detailed guidance on integrating trained models into business processes, consult the Predictive Intelligence Workbench integration and customization documentation.

Use machine learning to optimize your business processes. You can train and implement ITSM Predictive Intelligence Workbench use case models to augment your existing application workflows.

Important:

Starting with the yokohama release, the ITSM Predictive Intelligence Workbench will reach its end of life (EOL) as part of its deprecation process. It will be completely deprecated and no longer deployed, enhanced, or supported from the yokohama release. To get the latest experience for this functionality, you must install the Task Intelligence for ITSM application (com.snc.itsm_ml_task) plugin. For more information, see Task Intelligence for ITSM

For details, see the Deprecation Process [KB0867184] article in the Now Support Knowledge Base.

Explore use case templates

Users with the piwb_admin or piwb_manager role can explore the prebuilt use case templates and create predictive machine learning models. To create a machine learning model, you first select a prebuilt use case template. Some of the prebuilt templates are guided and display the Guided Setup flag. These templates include a comprehensive setup process to help ease you through implementation. Non-guided templates display the Classic Setup flag.

Figure 1. ITSM Predictive Intelligence Workbench guided use case template

Figure 2. ITSM Predictive Intelligence Workbench classic setup use case template

Templates with available pretrained models accelerate your setup process, by providing a pre-generated model based on your data. When a template indicates Pretrained, this means you can go directly to the evaluation phase of the use case setup. If the pretrained model is acceptable, you can directly integrate it with your business processes. Otherwise, you can tune this model or create another model. You may change the name and description of the use case later. Pretrained models display the estimated percentage of your correctly predicted incidents.

Templates with available pretrained models also display the estimated percentage of the correctly predicted incidents. If the pretrained model is acceptable, you can directly integrate it with your business processes. Otherwise, you can tune this model or create another model. You may change the name and description of the use case later. Pretrained models display the estimated percentage of your correctly predicted incidents.

Figure 3. ITSM Predictive Intelligence Workbench pretrained model use case template

Non-guided, Classic Setup templates provide links to relevant Predictive Intelligence Workbench product documentation or link to the ServiceNow platform Predictive Intelligence application with the Take me therebutton.

Figure 4. ITSM Predictive Intelligence Workbench links to product documentation and Predictive Intelligence product documentation

Use case creation phases

Creating a predictive machine learning model involves several phases. After you create and train your model you need to evaluate and tune it, test its prediction results, and then integrate it with your business process. Use case model creation phases include:

Create and train models: Define parameters to create a model that you will train based on your unique data. It is common to create multiple models in this phase. You will tune and refine your models by defining the right combination of coverage and precision to use.
Test your models: Get prediction results from your models to decide which one is best to integrate with your business process. To see if a model returns a correct result, you can use either the single or batch testing process.
Integrate the best model: Deploy the best model into your business process. After you determine which model returns the best, correct result, integrate it into production.
Note:
For details regarding trained use case integration implementation, refer to Predictive Intelligence Workbench integration and customization.