2026.2 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.

2026.2.0 Release Overview

Pre-Release

These release notes are being provided in advance of the release, for your preparation and evaluation of the upcoming version. This version will be released for Cloud Availability shortly.

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

Release 2026.2.0 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 2026.2.0 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
Unified Narrative Administration
NLQ Assistant – Natural Analytical Responses (Preview)

Unified Narrative Administration

All narrative-generation capabilities have been consolidated under a single Narrative setting, giving admins a clear, consistent way to control and monitor narrative usage across the product.

A single Narrative toggle in Sisense Intelligence Admin now controls:

• Dashboard/widget Narratives

• Narrative API usage (unified narrative REST API)

Cloud Linked Features Dependency

If Cloud Linked Features (CLF) is OFF, Narratives are disabled, even if the Narrative toggle is ON or covered by license. In this case, users see a notification indicating that Cloud Linked Features must be enabled for Narratives (and other AI capabilities) to work. This applies to:

• Dashboard/widget narratives

• Assistant-generated narratives

• Narrative API calls

Admin Impact and Configuration

To enable Narratives, ensure:

• Cloud Linked Features is turned ON, and

• Narrative is enabled under Sisense Intelligence > Feature Management.

Disabling the Narrative toggle:

• Hides or disables the Generate Narrative button in the Assistant.

• Disables narrative generation in dashboards/widgets and via the narrative API.

NLQ Assistant – Natural Analytical Responses (Preview)

The Sisense assistant now answers analytical questions with natural-language responses that include exact data values and concise insights, while continuing to generate the relevant visualization.

• Natural-language analytical answers – Quantitative questions now return direct answers with exact values plus a focused insight (for example: trend, ranking, or share of total).

• Question-aware responses – The assistant adapts responses by query type, including:

  • Quantitative – direct answer + insight

  • Creation requests – creates the requested visualization and confirms the action with a brief insight

  • Comparison – includes values, deltas, and percentages

  • Metadata / structural – returns text-only field or definition information

  • Ambiguous questions – asks clarifying follow-up questions with suggested options

  • Forecasting / trends – summarizes projected direction alongside the visualization

  • No data found – returns a clear text explanation without showing an empty visualization

• Shared filter context – Assistant responses and generated charts now respect active dashboard filters, ensuring consistent results across dashboards, NLQ, and assistant interactions.

Preview Scope

This release covers natural-language responses for core analytical question types and filter synchronization. Governance controls and complex explanatory “why” questions remain out of scope for this phase.

What's Improved

Analytical Engine

• Easier Analytical Engine Updates: You can now apply changes to Analytical Engine settings for ElastiCube and B2D models using the Publish Semantics option. This enables you to update model behavior without requiring a full data build.

Compose SDK Mode

• Compose SDK Mode now supports embedding code at the widget level.

Sisense Intelligence

• When using the assistant with Azure GPT 4.1, chart generation speed has been improved.

• The assistant now supports shared formulas.

WAT

• The URL length for the Jump to Dashboard in the WAT flow has been optimized by adding a mechanism to compress the 'filter' parameter in the URL. This reduces the number of cases where the URL is too large for the nginx server to process. (In some cases, it may be necessary to adjust the nginx server default parameters).

What's Fixed

Access Control

• Tightened access control for the /storage/fs/validate_file endpoint so that only users with permission to upload data sources (Data Designer and higher roles) can access it; lower-privilege roles now correctly receive HTTP 403.

Add-ons

• Previously, when using JTD, increasing the popup window size to 130% or larger sometimes caused the popup to be truncated, cutting off the dashboard header, navigation buttons, and scrollbar. This has now been fixed, and navigation buttons, windows, and menu bars remain visible in JTD popup windows, even when resized above 130%.

• Fixed an issue where enabling or disabling add‑ons (plugins) on self‑hosted Sisense deployments running with a non‑default Linux user ID (non‑UID 1000) could fail with a Permission denied (EACCES) error on licensed plugin files (for example, licensed-plugins/<plugin>/plugin.json). As a result, plugin state changes (enable/disable) appeared to revert and could not be applied.

  File ownership for licensed plugin files is now aligned with the configured deployment UID (securityContext runAsUser/fsGroup), so plugin state changes work as expected on deployments using custom UIDs.

APIs

• Previously, updating data security rules via REST API 1.0 PUT /elasticubes/live/datasecurity/{dataSecurityId} stored the partyId as a string instead of an ObjectId. This caused data integrity issues and could break downstream processes expecting the correct data type. The API now correctly stores partyId as an ObjectId.

• Previously, the PDF preview layout was not saved if the dashboard contained a widget added via POST /api/v1/dashboards/${dashboardId}/widgets/. The issue only occurred via API, not if the widget was added using the UI. This has now been fixed, and the PDF preview layout is saved when using the API as well.

Build to Destination

• Previously, the storage size shown for Snowflake Build-to-Destination ElastiCubes was much lower than the actual ElastiCube size. This has been fixed and Snowflake B2D cubes now display an accurate approximation of their actual storage footprint in Snowflake.

Data Models

• Previously, importing or exporting models with a high number of Data Security rules (~2,700+) caused a pod to crash. This was due to the system attempting to encrypt each security rule member individually, leading to TCP port exhaustion and a loss of connectivity for critical services. The encryption logic in the model import/export flow has been refactored to use a bulk encryption API with chunking (processing items in batches of up to 200). This dramatically reduces the number of concurrent HTTP requests, preventing port exhaustion and ensuring stability during large model operations.

Export to PNG

• Previously the user profile language was used for dashboard PNG exports, ignoring the lang URL parameter. This has now been fixed and PNG exports respect the lang URL parameter.

Filters

• Previously, selecting N/A plus other values in a widget sometimes caused expanded filters to display incorrect results. This has been fixed and filters work as expected.

• Previously, charts such as bar charts or column charts could not be filtered by the columns/bars with the following condition: Days granularity in the categories and Days in Date dashboard filters. This has been fixed and interactive filtering now works as expected.

Git

• Previously, in some cases, syncing dashboards via Git that contain rich text widgets, Git showed uncommitted changes even when no edits were made. This has been fixed and Git only shows actual uncommitted changes.

Outer Joins

• Previously, applying an include filter to show only specific members would still return unexpected NULL entries in the results. This occurred in data models that use a combination of outer and inner joins between tables, where the filtered field was connected through multiple levels of joins. This has been fixed and queries now correctly exclude NULL entries and return only the filtered members.

• Previously, when using Outer Join with Redshift, queries sometimes failed with a "column reference is ambiguous" database error. This occurred when the query included columns from different tables that had the same name but different casing (e.g., Country_id and country_id). This issue has been resolved and columns with identical names but different casing are properly disambiguated.

• With this fix, perspectives now correctly persist and apply the join type from the root model, ensuring LEFT JOIN behavior is honored in both the UI and the underlying queries, and eliminating the need to manually re‑define join types in each perspective.

• Previously, in ElastiCube models, filtering for NULLs on the null-producing side of an outer join resulted in zero rows. This has been fixed and the underlying SQL takes into account NULL values produced by the outer join as well as NULL values in the original tables.

• Previously, when applying a top or bottom ranking filter on a dimension that uses an outer join, rows where the dimension value is N/A were incorrectly excluded from the results. This issue has been resolved and N/A dimension values are now correctly included in Top/Bottom ranking results when using outer joins.

Perspectives

• Previously, in the Sisense semantic model, perspectives did not always correctly inherit LEFT JOIN relationships defined in the root model. In some cases, joins that were configured and displayed as LEFT JOIN in the UI were executed as INNER JOIN in the generated SQL for perspective‑based widgets, causing inconsistent and inaccurate results. This has been fixed and Perspectives apply the join type from the root model, ensuring LEFT JOINs execute as expected.

Report Manager

• Previously, in some cases, when exporting a report from Report Manager in a live model, the export failed due to a broken retry mechanism. This has been fixed by modifying the retry behavior for report exports.

• Previously, in some cases, scheduled emails sent by Report Manager to external recipients failed with a vague error description in the UI that could not be used to troubleshoot. This has been fixed and the UI now displays a clear error message.

• Previously, in some cases, Report Manager UI indicated a scheduled report was sent even though some recipients did not receive the report. This has been fixed and the UI now accurately reflects actual report delivery.

Search

• Previously, when performing a search in the left navigation panel, the UI occasionally froze for up to 10 seconds. This has been fixed by moving search functionality to the server, which subsequently frees up client resources.

Single Sign On (SSO)

• Previously, users were sometimes unable to authenticate via SSO due to a malformed redirect URL (//openid) caused by the browser’s cache. This has been fixed by correcting the way URLs are constructed, thereby resolving failed logins without requiring users to clear their browser cache.

Sisense Intelligence

• Previously, when the assistant generated a new visual or metric, you could only see the updated/new visual by clicking outside of the popup window and reopening the assistant. This has been fixed, and the visual is now updated in the existing widget.

User Parameters

• Previously, when an admin manually created a user with an email address that contained an invisible (non-printing) character and the same user logged in via SSO without using the invisible character, Sisense treated them as 2 separate accounts. This led to duplicate accounts that appeared to have the same email address, causing issues with dashboard access and permissions. This has now been fixed by adding validation that blocks the use of email addresses containing invisible characters both in the Sisense UI and in the backend.

Widgets

• Previously, in Pivot2 widgets, null values were sometimes displayed as 0 instead of N/A after changing the number of decimal places displayed in a column, leading to null values being interpreted as actual data. This has been fixed and Pivot2 widgets display null values as N/A.

• Previously, in pivot widgets with many columns, the end of the last column was occasionally cut off, making the resizing handles and sorting button inaccessible. This has been fixed by updating the widget UI and scrolling logic so that all columns and buttons are visible regardless of the number of columns.

• Previously, when clicking cancel while a widget is loading, the widget’s toolbar disappears and users cannot open or edit the widget. This has been fixed and the toolbar no longer disappears.

• Previously, Pivot widgets with outer joins encountered a query error after an upgrade if joined tables contained columns with identical names. This has been fixed and Pivot widgets now function correctly, even when joined tables share column names.

• Previously, in some cases, the scroll bar was hidden in table widgets with many columns, preventing users from scrolling through the table. This has now been fixed, and the scroll bar appears as expected regardless of screen resolution.

Deprecations

• Nginx ingress is being deprecated. It is being replaced with Nginx Gateway Fabric (NGF).

• The following old Narration endpoints are being deprecated:

  • /ai/nlg/queryResult

  • /ai/describe/query