Chat Artifacts
Analytics Chat now keeps model branches, Workbooks, Dashboards and saved explorations connected to the conversation. Return to a chat to pick up both the discussion and the artifacts you created or linked along the way.
You can now isolate data model changes to individual chats and work on multiple changes in parallel. Switch between chats without mixing your changes or losing the context behind each task. The Artifacts tab lets you preview saved work and inspect file changes directly from the conversation.
This is another step toward the agentic analytics loops we described earlier: conversations that carry their context and artifacts as multiple analytics tasks move forward in parallel.
Other improvements and fixes:
Workbooks
- Added filtering support for bins and value groups in Workbooks
- Added Fill in Missing Rows support for empty bins and value groups
AI
- MCP clients can now query development branches and test their model, skills and rules before publishing
- Added branch management support for MCP clients, including committing, publishing and discarding model changes
Embed
- Added PNG and PDF download support for entire Dashboards in embeds with exports enabled
- You can now disable AI for individual embed customers while keeping manual authoring available
- Added language and time zone settings for individual embed sessions
Integrations
Fixed incorrect pivot totals and totals disappearing after refresh in the Google Sheets and Excel add-ons.
Modeling
- Added support for importing models from an uploaded dbt manifest, so Cube can use your CI output without access to the dbt repository
- Fixed rolling-window queries on Snowflake failing without a date range when using standard time granularities
API
- Added Dashboard PNG and PDF export support to the public REST API for your own reporting workflows
- Fixed SQL pushdown for views spanning multiple data sources when a query uses members from just one source