Published Dashboards can now include a built-in Dashboard Agent so viewers can ask questions about what they're looking at. A chat bubble appears in the bottom-right corner — click it to open a side panel and the dashboard reflows alongside it.
The agent is read-only and grounded in what the viewer sees: it knows the dashboard's widgets, the charts they render, and the viewer's active filters and time grains, so answers always reflect the current view. Ask it to summarize the dashboard, explain a trend, or compare widgets — much like Analytics Chat, but for consumers rather than editors. Conversations persist in the URL, so a refresh restores the thread right where it left off.
The Map chart now supports a Region mode that plots query results as a choropleth over polygon GeoJSON. Choose a built-in source — World Countries or US States — or point to any custom HTTPS GeoJSON URL, then join it to your data on a region property. Cube auto-detects the join fields and color-scales regions by your measure, with optional shading for unmatched regions.
loadQueryResults MCP tool to paginate through large query resultsllm: auto to automatically select a modelpublic: falseYou can now share an Analytics Chat thread with teammates as a read-only conversation. Click Share in the chat header to grant view access to individual users, a user group, or flip General access to Organization to make the chat visible to everyone. Use Copy link next to Share to grab a direct URL once access is set up.
Recipients open the chat at the same URL as the owner and can scroll through the full conversation, expand the agent's reasoning and tool calls, and explore the charts and tables it produced — but they can't send new messages. Only the owner can continue the thread, and the header shows a "Shared by …" label so viewers always know whose conversation they're reading. If the owner asks more questions later, those messages and the agent's replies show up for viewers on the next load. Recipients still need access to the deployment and the underlying data the agent used; otherwise they'll see a "Chat not available" page.
End users can now share Workbooks they create in Creator Mode with each other, so saved explorations can be collaborated on from inside your embedded app.
Views now support default_filters so you can apply filters automatically to every query against a view.
The new Access Policies viewer surfaces every access policy defined in your data model — row-level filters, member-level restrictions, and member masking — broken down by the user groups they apply to. Audit who can see which cubes and views, and how each policy is composed, without grepping through cube files or running test queries.
Open it from the Model module by clicking Access Policies in the sub-sidebar. The viewer reflects whichever branch and build you are currently on, so policies you are editing in development mode appear alongside what is live in production.
We've rebuilt our dark theme from the ground up to improve its look and feel — with refined contrast, more readable typography, and polished surface styling across every Cube surface. Switch between light and dark themes from Preferences.
generate-session API now accepts a userProfile payload to set display name and avatar for external usersYou can now duplicate an exploration with Save as in the Google Sheets and Excel add-ons.
You can now organize your semantic model into view groups. Group related views by domain or purpose — Sales, Marketing, Customers, Inventory — and give each group a title and description so data consumers, AI agents, and embedded apps can discover the right dataset faster.
View groups are defined directly in the data model with the new view_groups top-level block and are surfaced everywhere views are picked across Cube — including the semantic view picker in Workbooks, Explore, and the Cube REST API meta response. The schema editor in the new IDE understands the new syntax, with autocomplete and validation for view_groups and the views array.
agents/ scaffold to make it easier to configure rules and certified queriesgenerate-session APIYou can now bring your own visual identity to Cube with custom chart palettes and dashboard themes. Define palettes once and apply them to any chart so visualizations stay consistent and on-brand across your workspace.
Dashboard themes let you configure, save, and reuse a complete look and feel for Dashboards — including colors, typography, and surface styling. Together, palettes and themes give you the flexibility to tailor the Cube experience to your organization's brand and pair especially well with our embedded analytics offerings.
You can now download Dashboards as PNG or PDF.
Creator Mode lets you embed the full Cube application into your product, so your users can build and modify their own dashboards without leaving your app. Instead of embedding a single dashboard or chat interface, you give users the complete Cube experience — creating and updating workbooks, publishing them as dashboards, and sharing them with their team.
Creator Mode is isolated on a per-tenant basis and supports additional layers such as groups inside tenants, so you can model your customers and their teams cleanly. It also supports customization at both the dashboard and application levels, so you can bring your own design system into Cube.
We're continuing to invest in Creator Mode and plan to add interactivity via events for building interactive experiences, as well as support for scheduled notifications.
The new filters are now available in the Google Sheets and Excel add-ons.
View descriptions are now surfaced in the Data Pane.
You can now drag and drop Workbooks and Dashboards into folders in the Workspace.
Anyone in your Slack workspace can now send a direct message to the Cube app and get AI-powered answers from the Analytics Chat agent — without leaving Slack. The agent responds in a Slack thread, and follow-up questions in the same thread keep the conversation contextual.
Cube matches Slack users by email address, so no additional account setup is required for your team.
To get started, a workspace admin connects Slack from Settings → Integrations → Slack and selects a deployment and agent to power responses. Learn more.
The filtering experience across Explore, Workbooks, and Dashboards has been updated. The point-and-click flow is faster in Explore and Workbooks, and you can now change filter operators in a single click.
Filters now support complex logical groups (OR and AND). You can create and edit these groups directly in the UI. All filters are rendered to Semantic SQL, so they work seamlessly with AI agents — both humans and agents see the same source of truth.
Dashboard filters have also been updated with the ability to change filter operators directly on Dashboards.
The documentation website has been updated with improved structure, search, readability, and mobile support. It also includes an AI-powered chat — ask the assistant your documentation questions directly from the docs.
With workbook versions, users can roll back changes to their workbooks and restore the state of the workbook as it was published at a given point. That is useful if you accidentally made changes to the published dashboard and you need to roll back to the published state.
With this feature, users can send multiple messages to chat that will be queued together and sent automatically, one after the other. Users can also remove them and edit the content while the message is in the queue. That helps to provide more context and more tasks to AI, and let it run faster for a longer period of time.
meta.ai_context field. Learn moreCalculated fields let you define query-level calculations — such as custom measures or dimensions — directly from the Explore and Workbook interfaces. Under the hood, they are compiled into Semantic SQL and executed on the backend, so they work seamlessly with the rest of your data model.
This release introduces common measure calculations (e.g., difference, percentage of total) and quick aggregates for dimensions. Filtered measures and additional calculation types are coming in future releases.
AI agents can create calculated fields as well, enabling end-to-end analytics workflows powered by natural language.