- 10 Apr 2025
- 13 Minutes to read
- Print
- PDF
Predictive AI
- Updated on 10 Apr 2025
- 13 Minutes to read
- Print
- PDF
Predictive AI is a type of AI that uses historical data, statistical algorithms, and machine learning techniques to forecast future outcomes. Its main goal is to make predictions about events based on patterns found in existing data.
Prerequisites
Ensure to enable Copilot within the Instance on the Application Custom Script.
Enable Copilot on AI/ML of the Service Management Application. For more information, refer AI/ML.
Configure the AI Suggestions Popover Form and add it to the Form Action in Manage Incidents and Manage Requests pages.
Service Desk Intelligence
In addition to automatically resolving some selected IT service issues described in Conversational Copilot, Copilot also has the functionality of AI-driven automatic Categorization and Classification of all Incidents and Service Requests. This makes an analyst’s job fast and simple.
The predictions made by an AI model reflect past data. Based on input in the Symptom and Description fields, Service Desk Intelligence scans the available history of the records to identify similar records. Further, the Service Desk Intelligence analyzes human responses to similar records in the past and provides suggestions. Analysts are enabled to view and apply the AI suggestions provided by the Service Desk Intelligence. If records exist with proper data, the field predictions are in conjunction with that.
Use Case
Predictive AI enhances IT Service Management (ITSM) by proactively addressing potential issues and streamlining operations. A notable use case is its application in Service Desk Intelligence.
Use Case | Solution |
Novatech’s IT service desk receives a high volume of incident and service request tickets daily. Efficiently categorizing, prioritizing, and routing these tickets is crucial to meet Service Level Agreements (SLAs) and ensure prompt resolutions. | Predictive AI analyzes incoming tickets by examining their symptoms and descriptions. It compares this information against historical data to identify patterns and similarities. Based on this analysis, the system suggests appropriate categories and classifications for each ticket, reducing manual effort and potential errors. The system provides predictions for various ticket fields to the analyst to whom the issue is assi including:
These predictions assist analysts in making informed decisions quickly. |
Apply AI Suggestions
Copilot’s Service Desk Intelligence efficiently identifies the correct resolver group to address an issue by analyzing its specifics and matching the skillsets and expertise of the available support teams. Leveraging historical data and predictive analysis, copilot directs the issue to the analyst expertized in resolving it, enhancing resolution speed and accuracy. This capability optimizes resource allocation and improves overall service quality.
To Apply AI suggestions on an Incident or Request, perform the following steps:
Log In to the application as an Analyst.
Select Service Management.
Navigate to Incident > Manage Incidents.
Manage Incident - List page is displayed.Figure: Manage Incident - List page
Click Incident ID hyperlink to view and apply AI suggestions.
Note
Ensure to select the Incident ID that meet the condition as defined on the AI Suggestions Popover Form.
Example: If Status is New, In-Progress
By default, AI Suggestions popover form is auto displayed when an Incident is opened.
Validate the values and click Apply to insert the values in the fields.Figure: AI Suggestions
For more information, refer the following Field Description.Field
Description
Workgroup
A group entitled to work or resolve the selected Incident / Request.
Example: Helpdesk, Application Support, Infra Support Team etc.Impact
Measure of the effect of an incident/Request on the business processes. Based on the number of affected users and the business criticality. An Impact is determined.
Example: If Impact is High it implies that a large number of users are affected impacting business activities.
Urgency
Measure of how quickly an incident must be resolved and resolved to prevent further business impact.
Example: If Urgency is High it implies that Incident / Request must be resolved or completed.
Priority
Based on the measure of Impact and Urgency on a Request / Incident, Priority value is determined.
Example: If Impact and Urgency are High, then the Priority can be P1, P2, or P3
Category
Categorize the Incident / Request to a category group according to certain shared characteristics such as the nature of the issue, affected systems, or the type of service impacted.
Example: Infrastructure, Administration, etc.Classification
Classify the Incident / Request to a group that share similar occurrences based certain characteristics, like the type of issue, its consequences, and the systems or services that are impacted.
Example: Hardware, Access, etc.Analyst
Name of the Analyst to whom the Incident or Request will be assigned to work upon.
Disable a suggestion by deactivating the control for the field that is not required. An additional field Justification for not using the <filed_name> is displayed. However, providing justification is optional.
Figure: Disabled fields on AI SuggestionsClick Apply to insert the values in the fields and return to the Record details page.
A success message is displayed as Field suggestions from AI are applied.Figure: AI suggestion Success Message
AI Suggestions Form
The AI Suggestions is a Popover Form configured in the Form Designer of Service Management Application and used as a Form Action on the Manage Incidents Form. This is one of OOB forms provided within the application.
Access AI Suggestions from the Record
Apart of auto-display of AI Suggestion on an Incident / Request, this form is also accessed from Record’s Form Action. To access AI Suggestion from Form Action, perform the following steps:
Log In to the Application as an Analyst.
Select Service Management Application.
Navigate to Incident > Manage Incidents.
Manage Incident List - page is displayed.Click Incident ID hyperlink to view the record.
Note
Ensure to select the Record for which AI Suggestion condition is met, otherwise you will not able able to view the AI Suggestion popover form.
On the top right of the Incident click vertical ellipsis and select AI Suggestion.
Figure: Incident Details pageAI Suggestion page is displayed.
Make the required changes and click Apply.Figure: AI Suggestions Popover Form
Note
AI Suggestions popover form is displayed when an Incident is opened for the Analyst whom the Incident is assigned.
SLA Management and Risk of Escalation
In a high-demand IT environment, ensuring adherence to Service Level Agreements (SLAs) is critical for maintaining operational efficiency and user satisfaction. Predictive AI brings a proactive edge to SLA Management by forecasting the likelihood that an incident will breach its SLA commitment, specifically the Response and Resolution SLAs.
Use Case
This use case demonstrates how Predictive AI helps prevent SLA breaches by identifying high-risk incidents in real time. By forecasting the likelihood of Response and Resolution SLA violations, it enables timely escalations and smarter resource allocation.
Use Case | Solution |
An enterprise IT service desk receives a critical incident, The ticket is automatically assigned a Response SLA of 30 minutes and a Resolution SLA of 4 hours. During this time, the service desk is experiencing high ticket volume and limited analyst availability due to a shift change, increasing the likelihood of SLA breaches. |
|
Leveraging historical data and real-time incident information, Predictive AI continuously analyzes patterns and current conditions to determine the risk level associated with each incident.
Response SLA Prediction: Evaluates the probability that the incident will not receive an initial response within the agreed timeframe.
Resolution SLA Prediction: Assesses the likelihood that the incident will not be fully resolved before the resolution deadline.
A higher risk level indicates an increased chance of SLA violation, allowing service teams to take proactive measures such as reassigning tickets, escalating issues, or reallocating resources before a breach occurs.
Complementing these predictions is the Risk of Escalation bar, a visual indicator that reflects the probability of an incident requiring escalation to meet SLA deadlines. This bar is calculated based on both current incident conditions and historical resolution patterns. It offers a clear, real-time view of escalation likelihood, empowering IT teams to prioritize efforts, manage workloads effectively, and ensure timely resolutions.
Response SLA
The Risk of SLA breach bar shows the likelihood that the Incident will not meet the Response Service Level Agreement (SLA) deadlines. This assessment is based on analyzing current conditions and historical data patterns. A higher risk level indicates a greater chance of failing to respond within the agreed timeframe, enabling proactive measures to prevent SLA breaches and maintain service quality.
Note
Risk of Response SLA breach bar is displayed when the following conditions are met:
Status of the Incident must not be closed or cancelled.
Incident with unrecorded Response SLA.
To view the Risk of Response SLA breach for an Incident, perform the following steps:
Log In to the Application as an Analyst.
Select Service Management Application.
Navigating to Incident > Manage Incidents.
Manage Incident - List page is displayed.Click Incident ID hyperlink to view the Incident details.
Navigate to Response SLA widget.
Observe the Risk of Breach as a bar.Figure: Response SLA Widget
Hover over the Risk of Response SLA Bar, to view details in a tooltip.
Figure: Risk of Response SLA - tooltip
Resolution SLA
Risk of Resolution SLA breach reflects the likelihood that an incident will not be resolved within the agreed Resolution SLA timeframe. This assessment is based on analyzing both current conditions and historical data trends. By evaluating patterns from past incidents and the present situation, this metric helps predict whether the incident resolution is at risk of exceeding the set SLA limits, enabling proactive measures to avoid potential breaches.
Note
Risk of Resolution SLA breach is displayed when the following conditions are met.
Status of the Incident must not be closed or cancelled.
Incident with unrecorded Resolution SLA.
To view the Risk of Resolution SLA breach for an Incident, perform the following steps:
Log In to the Application as an Analyst.
Select Service Management Application.
Navigating to Incident > Manage Incidents.
Manage Incident - List page is displayed.Click Incident ID hyperlink to view the Incident details.
Navigate to Response SLA widget.
Observe the Risk of Breach as a bar.Figure: Risk of Resolution SLA Breach
Hover over the Risk of Response SLA Bar, to view details in a tooltip.
Figure: Resolution SLA tooltip
Risk of Escalation
The Risk of Escalation bar visually represents the probability that an incident will need to be escalated to meet SLA deadlines, using current and historical data patterns for its calculation. By analyzing trends and past incidents, the bar provides a clear, real-time indication of how likely it is that the issue will require higher-level attention to ensure timely resolution. This helps in proactively managing incidents and allocating resources effectively to avoid SLA breaches.
To view the Risk of Escalation for an Incident, perform the following steps:
Log In to the Application as an Analyst.
Select Service Management Application.
Navigating to Incident > Manage Incidents.
Manage Incident - List page is displayed.Click Incident ID hyperlink to view the Incident details.
Risk of Escalation is displayed on the top of the Incident details screen.
Figure: Risk of Escalation bar
Hover over the Risk of Escalation bar to view more details about it in a tooltip.
Figure: Risk of Escalation tooltip
Change Approval Risk Analysis
Copilot provides intelligent risk advisory for planned changes by providing the risk and success rate of proposed changes on the basis of historical data and predictive analysis. When an approver reviews a change request, copilot assesses previous change outcome and identifies potential risk associated with the change request. This insight helps the approver to make informed decisions, enhancing the reliability and success of change implementations.
Use Case
This use case demonstrates how Predictive AI helps Change Approvers to review the insights provided by the copilot to make the appropriate decision to Approve or Reject a Change Request.
Use Case | Solution |
A DevOps engineer submits a change request to upgrade a core database server from version 12.1 to 13.0 in the production environment. The request includes technical documentation and a proposed rollback plan. | The change approver (a CAB member) opens the request in the Service Management Application, which is integrated with Change Approve Copilot. Copilot analyzes historical data from the change management system, identifying previous changes of similar type (example: database upgrades). It compares environment (production), components affected, change initiator, timing (example: end-of-quarter), and whether incidents followed those changes. Based on these analysis Copilot assigns success probability (example: 92%) to this change. The approver reviews the insights and decides to approve the change. |
To view likelihood of change success, perform the following steps:
Log in to the Application as an Approver.
Select Service Management.
Navigate to Change > Approve Change Requests.
Approve Change Request - List page is displayed.Click Change ID hyperlink to approve a CR.
Ensure to click on CR ID that is Awaiting Approval.In the Change Request Details page click Approve.
Figure: Change Request Details
A new pop-up window is displayed. It shows the Likelihood of change success represented in the form of a bar along with percentage rate of success and a note whether it is safe or not to approve the Change Request.
Figure: Change Approval Risk Analysis barNote
The bar indicates the chance of successfully implementing the change, considering current settings, past results, and predictive models. A higher percentage means a higher likelihood of success.
Hover on the predicted risk bar to view the tooltip.
Figure: Workflow Approval Change Risk tooltipClick Status column to select Reject or Approve.
Figure: Workflow ApprovalClick Submit after selecting the Status.
Change Risk Prediction
Change Risk level predicts the potential threat in implementing the change. A Low risk level suggests that the change carries minimal risk and is unlikely to cause significant issues. Conversely, a High risk level indicates that the change is associated with substantial risk and could lead to serious problems if not managed carefully. This assessment helps in making informed decisions and preparing appropriate mitigation strategies.
Note
Change Risk Prediction is visible to the Analyst persona.
To view Change Risk Prediction, perform the following steps:
Log in to the Application as an Analyst.
Select Service Management.
Navigate to Change > Manage Change Requests.
Mange Change Request - List page is displayed.Click CR ID and click General tab.
View the Predicted Risk under Impact section.Figure: Change Risk Prediction Bar
Hover over Risk bar to view the tooltip.
Figure: Predicted Change Risk
Change Type Prediction
Copilot predicts the change type for a suggested change record by analyzing historical data and patterns from previous changes. It evaluates various factors, such as the nature of the change, its potential impact, and past outcomes, to determine whether the suggested change should be classified as standard or normal. This prediction helps streamline the change management process by providing accurate categorization and ensuring appropriate handling of the change.
Following are the Change Types within the Change Request.
Standard Change: Are low Low Risk pre-authorized changes, such as installing additional hardware on a client PC or installation of OS patches.
For Standard Change, the CAB Approval is not required. After authorization, the CT is pushed for implementation. If the Change Type is selected as Standard in the Approval Tab, you will not see any CAB approval section.
Normal Change: The Authorizer and the CAB members must approve the CR to implement the change. If any of the CAB members reject the change based on the impact of cost, time or any other reason the CR cannot be implemented. After the all the approvals from the CAB members, a change is moved to the implementation stage.
To view Change Type Prediction, perform the following steps:
Log in to the Application as an Analyst.
Click Service Management.
Navigate to Change > Manage Change Requests.
Manage Change Record Details page is displayed.Click CR ID and select General tab to view the Change Type Prediction bar.
Figure: Change Type Prediction
Hover over the Change type to view the prediction tooltip.
Figure: Change Type Prediction tooltip