L2025.3 Release Notes

For a list of release dates and Sisense's end of support schedule, see Sisense Version Release and Support Schedule.

For information about the Sisense gradual rollout process, as well as an explanation of how the versions and release notes relate to content added during the rollout, see Sisense Gradual Rollout Process.

Versions Documented in these Release Notes

• L2025.3

• L2025.3 Service Pack 1

Release Overview

Release L2025.3 provides a number of new features, improvements, and fixes to Sisense for Linux.

What's New

The following table lists the high-level impact (or potential impact, if any) of new features, and how to handle it if upgrading to version L2025.3 or newer. Continue reading the Release Notes below the table for a detailed explanation of these features, as well as improvements and fixes.

Feature	Issues and Actions to Consider
Compose SDK Mode (Beta)	Compose SDK Mode is currently in Beta. Note the Known Issues in the documentation.
Display Name	Note the Known Issues below.
ElastiCube Cloud (Limited Availability)	The ElastiCube Cloud option should currently be considered by new users with small to mid-scale data usage, up to 80 GB. However, there is no concern of outgrowing this solution - while the initial limit will be 80GB, this will increase over time to meet any required size.
Infra	MongoDB has been upgraded to version 8. See Important Note above concerning upgrading Sisense. Added support for upgrading RKE and Kubernetes to version 1.33. Kubernetes is not upgraded by default during this update. To enable the upgrade, set the following parameter in your values.yaml file - update_k8s_version: true
Semantic Enrichment	See the Note below, and more information in the documentation.
Sidebar Navigation	N/A

Compose SDK Mode (Beta)

Compose SDK Mode is a powerful new dashboard feature designed to streamline embedding Sisense with Compose SDK.

• Renders dashboards in Sisense Fusion using Compose SDK for an accurate preview.

• Identifies compatibility issues and prevents incompatible modifications.

• Facilitates adaptation of existing dashboards to embedding with Compose SDK and creation of new, natively compliant dashboards.

Use Compose SDK Mode to build and embed better dashboards.

Display Name

Tables and columns in the data models will now differentiate between original, identity, and display names if enabled by an Administrator in the Admin tab > Feature Management page by turning on the Display Name toggle. This allows data designers to rename tables and columns without affecting their dependent assets, such as widgets or alerts.

• Original Name – The physical table/column name in the data source.

• Identity Name – The logical name in the data model schema, used as a table/column unique identifier. Changing it will cause existing queries relying on it to fail.

• Display Name – The customizable alias to be displayed in dashboards. Changing it will present an updated name to users in the Analytics page, but will not affect existing assets.

Changes to identity/display names will become effective after building/publishing the model.

Known Issues

• Metadata Aliasing – Search inside the data browser on the Analytics page returns the Identity name instead of the Display name when both the display name and metadata aliasing are applied.

• Smart field suggestions (both field names and table names) are not updated after publishing a live model. This functionality works correctly for ElastiCubes. Workaround: Manually trigger the POST suggestion/refresh API.

• Search in Analytics data browser – When searching for a table/column inside the data browser, the search returns results based on both Identity and Display names. For example: There is a “cost” column and a display name for it, “expense”. In the search field pop-up, this field will be returned if you search either for “expense” or for “cost”.

ElastiCube Cloud (Limited Availability)

ElastiCube Cloud is a modern, managed analytics solution. It represents a focused, scalable step toward the broader vision of a fully managed data warehouse (MDWH), while delivering immediate performance improvements and easier deployment. ElastiCube Cloud allows data modelers to use the Sisense cloud as a destination for their ElastiCubes. As this is managed on the Sisense cloud, there are no infrastructure and operation burdens on customers.

Infra

• MongoDB has been upgraded to version 8.

• Added support for upgrading RKE and Kubernetes to version 1.33.

  Note:

  Kubernetes is not upgraded by default during this update. To enable the upgrade, set the following parameter in your values.yaml file - update_k8s_version: true

Semantic Enrichment

Semantic Enrichment is a new feature that enables you to use Sisense Intelligence to automatically provide descriptions for your data models’ tables and columns. This assists data administrators in creating clear and informative text that can be utilized by both humans and the Sisense Intelligence assistant. The primary purpose of the enrichment is to enable the assistant to better understand the models and to improve the answers provided. The descriptions will cover key aspects of the data, such as its nature, its usage, optional formatting for textual IDs, and a summary of the data's time granularity and attributes.

1. In the Data tab, for the relevant data model, click > Semantic Enrichment.

   

2. If the Semantic Enrichment confirmation and note window is displayed, click Enrich Semantics.

   The Description fields (not including hidden tables/columns) are updated with the Semantic Enrichment automatically generated descriptions. Those descriptions are marked with the Sisense Intelligence icon .

   

Sidebar Navigation

The top navigation bar, including all the tabs and icons can now be moved to the sidebar. This feature is currently off by default. In order to use Sidebar Navigation, it must be enabled by an Admin, in the Admin tab, under App Configuration > Feature Management in the Management section.

What's Improved

Assistant

• Recommended questions in the assistant have been improved:

  • The assistant now suggests a wider variety of questions for deeper exploration.

  • The underlying API also supports a userPrompt field, allowing tailored instructions for more relevant recommendations.

    GET /ai/recommendations/query/{sourceId}/{numOfRecommendations}

Build

• A specialized retry mechanism has been implemented for dimension table corruption detection during cube builds.

  A new configuration property, “NumberOfBuildRetriesDimTable", has been added under the build's service configuration. Its value can be set to the range of 0 to 2 (default = 1).

  Whenever the build fails due to dimension table corruption detection and the configuration value is 1 or 2, a build retry will immediately be triggered.

  This specialized retry mechanism is an addition to the already existing "RetryBuild" and "RetryOnManualBuilds" features.

Narrative

• Widget narratives powered by Generative AI (Cloud only) have been improved. Cloud customers with the Narrative NLP package now get more insightful narratives using Generative AI. The previous version remains available for a limited time.

User Parameters

• The length limit for User Parameter values has been increased from 128 characters to 1024 characters.

What's Fixed

Access Control

• Previously, viewers incorrectly had access to the LLM configuration API /api/v2/settings/ai/llm. This has now been fixed and the access control is as expected.

Add-ons

• Report Manager - Previously, when Proxy URL was enabled, the Report Manager did not work properly due to an issue with the API URLs. Support for Proxy URL configuration of Sisense has now been fixed, and the Report Manager works as expected.

  Note: The Report Manager URL has now changed from /reportManager/main.html#/reports to /app/report-manager#/reports.

• Jump To Dashboard (JTD) - Previously, the “Jump to Dashboard Settings” menu option was sometimes available for unsupported widget types. This has now been fixed and the JTD settings are only available when relevant.

• Metadata Translation - Previously, when custom labels were defined for the X and Y axes on a column chart, the Metadata Translation add-on would ignore them and not translate them. This has now been fixed, and the custom labels are translated as expected.

Assistant

• An issue has been fixed where chart labels did not appear properly with multiple break-bys. The assistant has been improved to render labels on charts that include more than one break-by field for clearer and more accurate visuals.

• Stability improvements have been implemented for the Studio Assistant (Public Preview). Issues have been fixed for chart rendering, editing, missing data models, and model config updates to improve overall reliability.

• Previously, the analytics assistant was not available for WAT users. This has been fixed and is available for WAT users.

Browser Support

• The Feature Policy response header is deprecated and has been replaced with Permissions Policy.

Connectors

• Data source connection to MongoDB with the "Use Svr Connection" option is now working as expected. Previously, it failed with the following error: "Connection failed: Authentication."

Dashboard Co-Authoring

• Previously, when a dashboard had been shared before enabling co-authoring, then shared with co-owners after enabling the feature, and a different admin tried to access it with adminAccess, the dashboard did not load. This has now been fixed such that the dashboard loads and displays the data.

• The "Share the Assistant" toggle is now available in the drop-down menu for co-authored dashboards, and appears only when in shared mode, as intended.

Data

• In the list view of the Data page, when the cursor is on the data model title a pop-up window appears with the Analytical engine settings. However, it previously displayed “Compatibility mode” on data models that were created before upgrading to L2025.2 SP2, even if the user already modified it to “Analytical engine (Recommended)”. This has now been fixed and shows the correct state.

Data Security

• Previously, when Data Security was more complex, exporting it in the model may have resulted in a timeout, while still allowing the export to notify of a success. The timeout has now been extended from 10 seconds to 60 seconds. If it still fails, the entire export will fail and display a UI error accordingly.

Filters

• Previously, changing dashboard filter selection by an administrator in /adminAccess mode resulted in converting a multi-select filter into a single-select. This has now been fixed and works as expected.

• In some earlier versions, the blank/empty string values were not processed correctly in the legacy filter editor, being displayed as {“data“:””,”text”:””}. This has now been fixed, so that empty strings are presented as "(blank)" in the filters UI.

Forecast and Trend

• An issue has been fixed where special characters caused “Analyze It” trends to fail. A problem was resolved where values containing special characters (such as “&”) prevented trend analysis from working correctly in the assistant.

Git Integration

• Git projects are now included in the scope of backup and restore.

• Previously, when a live model was pulled from remote and the connection was updated, once the model was published it remained in a disconnected state. Now it will be presented in a published state.

• Hierarchy OIDs are now preserved during the pull to the target env. Previously, OIDs were regenerated, causing conflicts and disruptions in the data models. Customers may see a one-time unexpected change for existing hierarchies, namely a deleted + added file.

Query Plan Analyzer (QPA)

• When upgrading versions, a library required by QPA was missing. Because of that there was an internal error causing the graphs to appear empty. The library has been added to the installation and the graphs now appear correctly.

User Interface (UI)

• UI usability for the data theme has been improved for scroll bar and widget theme inheritance.

Widgets

• Previously, removing a value from a chart by clicking on the corresponding legend item did not remove its total value label. This has now been corrected such that when a value is removed, the related labels are removed as well.

• Previously, sorting date-time values from a Table widget column header was not applied correctly and instead resulted in a 'no sort' order. This has now been fixed, so that clicking on a Table column header sorts the values in ascending/descending order as expected.

• Previously, bar chart height was not always recalculated correctly after applying filters, leading to displaying the bars poorly laid-out tightly together at the top of the widget. This has now been fixed, ensuring that the bars are evenly spread across the chart area after a dashboard filter was applied.

• Previously, the scatter map did not return any results if empty strings were present in the data model and included in the query (e.g., due to applying filters). This has now been corrected by excluding the blank values from map queries to prevent them from breaking.

• Previously, when a semicolon (;) was used in locations, the scatter map did not load and returned the following error: "Mapbox error - Cannot read property '0' of undefined". This has now been fixed and works as expected.

• Previously, the sorting order of blocks representing values in the treemap widget was broken after changing the window size. This has now been fixed to maintain the correct sorting order of the values.

• Previously, long category titles in bar charts were not always truncated, affecting the chart display and readability. This has now been fixed to ensure that category titles do not obscure the chart area.

• Previously, clicking “Clear Selection” on a Pivot widget would sometimes reset the corresponding dashboard filter, without updating the widget itself. This has now been corrected to ensure that the selection in the widget is cleared as expected.

Deprecations

• The Quest add-on will be deprecated in the next release (L2025.4). If you rely on this feature or have concerns, contact your Customer Success Manager.

Coming Soon

Primary Filters (Preview)

• Primary filters can be configured by dashboard designers to dynamically restrict the values available in regular filters based on a selection in primary filters. This helps create a guided experience for end-users by eliminating irrelevant filter options.

  For example:

  When a user selects a particular store in primary filter ‘Store’, the regular filters ‘Product’ and ‘Brand’ will only present for selection products and brands, respectively, that are sold in that specific store.

  This feature is now available as a limited preview for testing. Administrators can enable primary filters by toggling on Primary Filters in Admin > Feature Management > Filters.

L2025.3 Service Pack 1

Cloud Availability

This release is currently only available to cloud customers. It will be released shortly for on-premise availability as well.

As the content is still in progress, it is recommended that you check back here occasionally for the latest updates.

The content below describes the improvements and bug fixes included in the August 2025, L2025.3 Service Pack 1 release.

For a list of release dates and Sisense's end of support schedule, see Sisense Version Release and Support Schedule.

For information about the Sisense gradual rollout process, as well as an explanation of how the versions and release notes relate to content added during the rollout, see Sisense Gradual Rollout Process.

What’s Improved

Connectors

The Analytical Engine (AE) is now fully supported for MySQL live models. The AE is disabled for MySQL by default. To enable it, toggle on the MySQL connector configuration parameter named enableAnalyticalEngine.value. For detailed instructions to toggle on the AE in your system, see Activating the Analytical Engine, in the “For Live models” section.

In an upcoming version, the AE will be enabled for MySQL by default.

Key Changes:

• All of the advantages of the Analytical Engine as described in Analytical Engine Overview - Key Differences and Advantages.

• New support in the MODE statistical function when using the AE in MySQL. This function was not supported in mySQL when using the legacy translator or AE until now.

• The Grand Total on Contribution results have been corrected in the AE.

Limitation:

• The PERCENTILE and QUARTILE statistical functions are currently still not supported in MySQL.

Studio Assistant (Now in Beta)

The new Studio Assistant is now in beta. This provides a generative AI-powered experience designed to accelerate analytics creation through natural language.

Later this year, this assistant experience will be expanded to power the Analytics Assistant, bringing the same level of intelligence and interactivity to dashboard viewers and designers.

The beta release of the Studio Assistant includes:

• Enhanced intent understanding: Vast improvements to the assistant’s ability to interpret ambiguous or incomplete prompts, leading to more accurate and relevant visualizations.

• Canvas mode: A new workspace for navigating across created artifacts: charts, dashboards, and data models. Easily jump between assets and view edit histories in context.

  • A drop-down menu enables you to switch between charts, dashboards, and data models without leaving the assistant interface.

  • The assistant now recognizes when edits are made to the same artifact type (i.e., dashboard, chart, etc.) and flags them clearly within the session flow.

• Thumbs up/down feedback: You can now rate the assistant’s responses with simple thumbs up/down telemetry, to help Sisense evaluate usefulness.

• Assorted UX improvements: Faster responses, smoother interactions, and contextual tooltips to guide user prompts.

• Improved chart creation experience:

  • Forecast & Trend Support: Ask for trends and forecasted values directly in natural language—no manual configuration required.

  • Full Data & Style Option Editing: The assistant now supports a comprehensive range of configuration commands to modify the chart with natural language.

• Improved dashboard creation experience:

  • In order for dashboards created with the assistant to persist, they must now be explicitly saved to the dashboard collection.

  • The new Dashboard Layout Mode enables you to drag-and-drop widgets for rearrangement and widget deletion.

• Improved data model creation experience

• Several bug fixes

What’s Fixed

Add-ons

• Accordion - Previously, when there was a large number of Accordion target dashboards (dashboards with the “_accrd” prefix), the “Select target” drop-down list did not display them all, with no ability to scroll to the undisplayed target dashboards. This has now been fixed and all target dashboards are displayed.

• Report Manager - The "Open Report Manager in Pop-Up" configuration option has been removed.

Assistant

• Previously, the "share the assistant" toggle was not present for dashboard co-owners. This has been fixed and appears for dashboard co-owners.

Build To Destination (B2D)

• Previously, in some cases, B2D endpoints failed with a 500 error. This has now been fixed, and they are working as expected.

Data Models

• Previously, the Refresh Tables Schema button refreshed the schema for only one table. This has been fixed, and now it does so for all of the tables.

Explanations

• Previously, filters were not applied correctly to explanation results. This has now been fixed and is working as expected.

Export to Excel V2

• Previously, the download menu contained a duplicate "Excel File" option. This has been removed.

WAT

• Export to PDF in WAT-based assets is now working as expected. Previously, in some cases it got stuck on the report preparation step and eventually failed with the following error: "Error during response error extracting: Failed to execute 'readAsText' on 'FileReader': parameter 1 is not of type 'Blob'."

Widgets

• Previously, clicking on the next or last page icons in page navigation resulted in pages that are indefinitely generated in the Table widget. They now point only to the next or last page.