--- sourceDocument: zurich Combined Product Files sourceDocumentLink: https://www.servicenow.com/docs/r/zurich/delta-xanadu-zurich Release : - zurich ft:locale : - en-US ft:publication_title : - zurich Combined Product Files ft:clusterId : - delta3zx bundleId : - delta3zx --- # Combined Predictive Intelligence release notes for upgrades from Xanadu to Zurich # Combined Predictive Intelligence release notes for upgrades from Xanadu to Zurich {#ariaid-title1} Release version: Zurich Updated October 8, 2026 ![](https://www.servicenow.com/docs/portal-asset/ico-clock) 7 minutes to read Summarize ![AI sparkle icon](https://servicenow.com/docs/portal-asset/ai-sparkle-icon) Summarized using AI This content was generated using new OpenAI-powered functionality. Results are provided on an as is basis and are not guaranteed to be accurate or complete. ## Summary of Combined Predictive Intelligence release notes for upgrades from Xanadu to Zurich This consolidated release note covers all updates, new features, changes, deprecations, and activation information for ServiceNow Predictive Intelligence from the Xanadu release through the Zurich release. Predictive Intelligence, formerly known as Agent Intelligence, has evolved with enhancements that improve model training, explainability, observability, and user experience. This summary helps ServiceNow customers prepare for upgrades by highlighting key improvements and important upgrade tasks. Show full answer Show less ## Important Upgrade Information * Before upgrading to Zurich, review and complete pre- and post-upgrade tasks to ensure smooth transition. * No specific release notes updates were noted for some intermediate releases (Xanadu, Yokohama, Zurich) in certain sections. ## New Features * **Model Explainability (Yokohama):** Adds a Feature Importance tab in Workflow Classification solutions to identify which classes most influence model predictions. * **Advanced Classification Options (Yokohama Patch 4 and Zurich):** New configurable features include: * Include only top N labels to focus the model on the most frequent labels. * Minimum records threshold per label to ensure data quality. * Option to remove the "others" label to reduce noise and improve accuracy. * Use of LightGBM algorithm for enhanced classification model training. * Ability to customize training with JSON-configured parameters. * **Observability Dashboard (Zurich):** Introduces a dedicated error logging table and dashboard to monitor and analyze prediction errors effectively. * **Coral Theme (Zurich):** Introduces a new default, brand-neutral theme for portals and mobile with an optional dark mode for better readability and reduced eye strain. * **Validation Logic (Zurich):** Ensures Predictive Intelligence has proper ACLs to access data tables, reducing training errors. ## Changes * Advanced classification options introduced in Xanadu Patch 9 are carried forward and enhanced in later releases. * Coral theme becomes default for new portals and interfaces starting in Zurich. * Validation logic added in Zurich helps prevent data access issues during model training. ## Deprecations and Removals * **Xanadu:** The dashboard for Solution Statistics is deprecated. Existing dashboards remain accessible but new customers will not have access to create them. * **Yokohama:** ITSM Predictive Intelligence Workbench is deprecated. Customers are recommended to use the Task Intelligence for ITSM application for similar functionality. * No features were removed in Zurich. ## Activation and Requirements * Predictive Intelligence is available as a ServiceNow AI Platform feature with appropriate plugin activation depending on release (from separate plugin activation in Xanadu and Yokohama to being active by default in Zurich). * No new additional requirements, browser requirements, or localization considerations were introduced through these releases. ## Accessibility and User Experience * The Coral theme introduced in Zurich includes a dark mode option for web and mobile to improve accessibility and readability. ## Highlights for ServiceNow Customers * Enhanced model explainability helps you interpret predictive outcomes more effectively. * Advanced configuration options give you greater control over classification model training and accuracy. * Observability features enable proactive monitoring and troubleshooting of prediction errors. * The Coral theme modernizes the user interface with improved aesthetics and accessibility. * Validation checks reduce training failures by ensuring proper data access permissions. * Be aware of deprecated dashboards and tools and plan to transition to recommended alternatives. Consolidated page of all release notes for Predictive Intelligence from Xanadu to Zurich. Predictive Intelligence was previously called Agent Intelligence. ## How to use this page To help you prepare for your upgrade, we have combined the cross-family Predictive Intelligence release notes onto one page. Read this summary of the new features, changes, and updated information for your product from Xanadu to Zurich. Tip: If there were no updates for a release notes section in a certain family release, we included a short note for your reference. For example, if a product did not have any updates in Tokyo, the row says "No updates for this release." ## Important information for upgrading Predictive Intelligence to Zurich Before you upgrade to Zurich, review these pre- and post-upgrade tasks and complete the tasks as needed. {#dfrn3-predictiveintelligence-summary__entry__2} | Release | Release notes | |-|-| | Xanadu | No updates for this release. | | Yokohama | No updates for this release. | | Zurich | No updates for this release. | [ ] ## New features Between your current release family and Zurich, new features were introduced for Predictive Intelligence. {#dfrn3-predictiveintelligence-summary__entry__10} | Release | Release notes | |-|-| | Xanadu | No updates for this release. | | Yokohama | [Model Explainability](https://www.servicenow.com/docs/access?context=predictive-intel-explainability&family=yokohama&ft:locale=en-US) : Learn which classes contribute most to your model's predictions by optionally adding Model Explainability to Workflow Classification solutions. Model Explainability provides a new tab labeled Feature Importance where you can run an analysis of each class's contribution to the overall prediction. [Leverage new advanced options for classification solutions](https://www.servicenow.com/docs/access?context=configuring-advanced-settings-ml-solutions&family=yokohama&ft:locale=en-US), from Yokohama Patch 4. : * [Configure include only top N labels](https://www.servicenow.com/docs/access?context=predictive-intel-only-top-n-labels&family=yokohama&ft:locale=en-US). Limit the classification model to use only the top most frequent labels. You can choose a number as the limit. * [Minimum records needed for label to include it](https://www.servicenow.com/docs/access?context=predictive-intel-minimum-records-needed-label&family=yokohama&ft:locale=en-US). Set a threshold for the minimum number of records a label must have in your dataset to be included in model training. * [Remove others label](https://www.servicenow.com/docs/access?context=predictive-intel-remove-others-label&family=yokohama&ft:locale=en-US). Reduce noise in the model and enhance predictive accuracy by removing records with the label "others" from training data. * [Use LightGBM algo for classification model training](https://www.servicenow.com/docs/access?context=predictive-intel-lightgbm-algo&family=yokohama&ft:locale=en-US). Enable the LightGBM (Light Gradient-Boosting Machine) algorithm for training classification models. * [Config parameters for model config in classification](https://www.servicenow.com/docs/access?context=predictive-intel-config-parameters-classification&family=yokohama&ft:locale=en-US). Customize training behavior by including a dictionary of parameters in the JSON format. | | Zurich | [Review any errors in predictions using the Observability Dashboard](https://www.servicenow.com/docs/access?context=prediction-errors-observability-dashboard&family=zurich&ft:locale=en-US) : Monitor errors using the Observability Dashboard which provides analytics derived from a new table. This table is dedicated to logging any errors that may occur during Predictive Intelligence predictions. [Leverage new advanced options for classification solutions](https://www.servicenow.com/docs/access?context=configuring-advanced-settings-ml-solutions&family=zurich&ft:locale=en-US) : Customize your classification models in Predictive Intelligence with the following advanced options. * [Configure include only top N labels](https://www.servicenow.com/docs/access?context=predictive-intel-only-top-n-labels&family=zurich&ft:locale=en-US): Limit the classification model to use only the top most frequent labels. You can choose a number as the limit. * [Minimum records needed for label to include it](https://www.servicenow.com/docs/access?context=predictive-intel-minimum-records-needed-label&family=zurich&ft:locale=en-US): Set a threshold for the minimum number of records a label must have in your dataset to be included in model training. * [Remove others label](https://www.servicenow.com/docs/access?context=predictive-intel-remove-others-label&family=zurich&ft:locale=en-US): Reduce noise in the model and enhance predictive accuracy by removing records with the label "others" from training data. * [Use LightGBM algo for classification model training](https://www.servicenow.com/docs/access?context=predictive-intel-lightgbm-algo&family=zurich&ft:locale=en-US): Enable the LightGBM (Light Gradient-Boosting Machine) algorithm for training classification models. * [Config parameters for model config in classification](https://www.servicenow.com/docs/access?context=predictive-intel-config-parameters-classification&family=zurich&ft:locale=en-US): Customize training behavior by including a dictionary of parameters in the JSON format. | [ ] ## Changes Between your current release family and Zurich, some changes were made to existing Predictive Intelligence features. {#dfrn3-predictiveintelligence-summary__entry__18} | Release | Release notes | |-|-| | Xanadu | [Leverage new advanced options for classification solutions](https://www.servicenow.com/docs/access?context=configuring-advanced-settings-ml-solutions&family=xanadu&ft:locale=en-US), from Xanadu Patch 9. : Five new advanced options are available for classification solutions, including new parameters and a new algorithm. * [Configure include only top N labels](https://www.servicenow.com/docs/access?context=predictive-intel-only-top-n-labels&family=xanadu&ft:locale=en-US). Limit the classification model to use only the top most frequent labels. You can choose a number as the limit. * [Minimum records needed for label to include it](https://www.servicenow.com/docs/access?context=predictive-intel-minimum-records-needed-label&family=xanadu&ft:locale=en-US). Set a threshold for the minimum number of records a label must have in your dataset to be included in model training. * [Remove others label](https://www.servicenow.com/docs/access?context=predictive-intel-remove-others-label&family=xanadu&ft:locale=en-US). Reduce noise in the model and enhance predictive accuracy by removing records with the label "others" from training data. * [Use LightGBM algo for classification model training](https://www.servicenow.com/docs/access?context=predictive-intel-lightgbm-algo&family=xanadu&ft:locale=en-US). Enable the LightGBM (Light Gradient-Boosting Machine) algorithm for training classification models. * [Config parameters for model config in classification](https://www.servicenow.com/docs/access?context=predictive-intel-config-parameters-classification&family=xanadu&ft:locale=en-US). Customize training behavior by including a dictionary of parameters in the JSON format. | | Yokohama | No updates for this release. | | Zurich | Coral theme : Coral is now the default theme for new portal, web, and mobile experiences with Next Experience or Core UI enabled. This theme provides a fresh look and feel, featuring brand-neutral illustrations to enhance your user experience. A dark theme option is available for web and mobile experiences. Validation logic ensures that Predictive Intelligence can access data tables : Reduce errors while training Predictive Intelligence models with the help of new validation logic. This validation checks whether your data tables have ACLs (Access Control Lists) granting access to Predictive Intelligence. | [ ] ## Removed Between your current release family and Zurich, some Predictive Intelligence features or functionality were removed. {#dfrn3-predictiveintelligence-summary__entry__26} | Release | Release notes | |-|-| | Xanadu | No updates for this release. | | Yokohama | No updates for this release. | | Zurich | No updates for this release. | [ ] ## Deprecations Between your current release family and Zurich, some Predictive Intelligence features or functionality were deprecated. {#dfrn3-predictiveintelligence-summary__entry__34} | Release | Release notes | |-|-| | Xanadu | The dashboard for Solution Statistics has been deprecated. Upgrading customers can continue to use their existing Solutions Statistics dashboards from the application menu. For new customers onboarding with the Xanadu release, the Solutions Statistics dashboard is not available. For more information, see [Review classification solution statistics](https://www.servicenow.com/docs/access?context=review-solution-statistics&family=xanadu&ft:locale=en-US). | | Yokohama | With the Yokohama release, ITSM Predictive Intelligence Workbench is deprecated and no longer supported. To obtain the latest experience for this functionality, install the Task Intelligence for ITSM application (com.snc.itsm_ml_task). For more information, see [ITSM Predictive Intelligence Workbench release notes](https://www.servicenow.com/docs/access?context=itsm-predictive-intelligence-workbench-rn&family=yokohama&ft:locale=en-US). | | Zurich | No updates for this release. | [ ] ## Activation information Review information on how to activate Predictive Intelligence. {#dfrn3-predictiveintelligence-summary__entry__42} | Release | Release notes | |-|-| | Xanadu | Predictive Intelligence is a ServiceNow AI Platform feature that is available with activation of the Predictive Intelligence Reports (com.glide.platform_ml_pa) plugin and its dependent Predictive Intelligence (com.glide.platform_ml) plugin. For details, see [Install Predictive Intelligence](https://www.servicenow.com/docs/access?context=install-predictive-intelligence&family=xanadu&ft:locale=en-US). | | Yokohama | Activation information : Predictive Intelligence is a ServiceNow AI Platform feature that is available with activation of the Predictive Intelligence plugin (com.glide.platform_ml). For details, see [Install Predictive Intelligence](https://www.servicenow.com/docs/access?context=install-predictive-intelligence&family=yokohama&ft:locale=en-US). | | Zurich | Activation information : Predictive Intelligence is a ServiceNow AI Platform feature that is active by default. | [ ] ## Additional requirements If any additional requirements were introduced or changed for Predictive Intelligence we have noted them here. {#dfrn3-predictiveintelligence-summary__entry__50} | Release | Release notes | |-|-| | Xanadu | No updates for this release. | | Yokohama | No updates for this release. | | Zurich | No updates for this release. | [ ] ## Browser requirements If any specific browser requirements were introduced or changed for Predictive Intelligence we have noted them here. {#dfrn3-predictiveintelligence-summary__entry__58} | Release | Release notes | |-|-| | Xanadu | No updates for this release. | | Yokohama | No updates for this release. | | Zurich | No updates for this release. | [ ] ## Accessibility information Review details on accessibility information for Predictive Intelligence, such as specific requirements or compliance levels. {#dfrn3-predictiveintelligence-summary__entry__66} | Release | Release notes | |-|-| | Xanadu | No updates for this release. | | Yokohama | No updates for this release. | | Zurich | Accessibility information : Dark theme : The new Coral theme includes a dark theme option for web and mobile experiences. This option is commonly used to alleviate eye strain and improve readability. | [ ] ## Localization information If there are specific localization considerations for Predictive Intelligence we have noted them here. {#dfrn3-predictiveintelligence-summary__entry__74} | Release | Release notes | |-|-| | Xanadu | No updates for this release. | | Yokohama | No updates for this release. | | Zurich | No updates for this release. | [ ] ## Highlight information If there are specific highlight considerations for Predictive Intelligence we have noted them here. {#dfrn3-predictiveintelligence-summary__entry__82} | Release | Release notes | |-|-| | Xanadu | * The dashboard for Prediction Results has been migrated to the Next Experience UI. Upgrading customers with existing Prediction Results dashboards can use either the Next Experience UI version or access the Core UI version from within the new dashboard. * The dashboard for Solution Statistics has been deprecated. Upgrading customers can continue to use their existing Solutions Statistics dashboards from the application menu. * New advanced options for classification solutions are available from Xanadu Patch 9. See [Predictive Intelligence](https://www.servicenow.com/docs/access?context=predictive-intelligence-landing&family=xanadu&ft:locale=en-US) for more information. | | Yokohama | * Learn which classes contribute the most to your model's predictions by adding Model Explainability to Workflow Classification solutions. * ITSM Predictive Intelligence Workbench is deprecated in the Yokohama release. * New advanced options for classification solutions are available from Yokohama Patch 4. See [Predictive Intelligence](https://www.servicenow.com/docs/access?context=predictive-intelligence-landing&family=yokohama&ft:locale=en-US) for more information. | | Zurich | * Predictive Intelligence errors are logged in a new, dedicated table. The PI - Observability Dashboard leverages this table to provide analytics on prediction errors. * New advanced options for Classification models are available, including new parameters and a new algorithm. * Validation logic ensures that Predictive Intelligence models have ACLs to access data tables. See [Predictive Intelligence](https://www.servicenow.com/docs/access?context=predictive-intelligence-landing&family=zurich&ft:locale=en-US) for more information. | [ ]