You can now filter with calculations post-query! This opens up a huge number of new use cases, filtering rows for visualization (ie trimming off partial moving averages), removing some rows (say hiding rows based on an IF calc), or even just simpler tables with calcs over a larger data set.

We're moving blobby into AWS Bedrock (powered by Claude), to reduce the surface are of where customer metadata is used and simplify subprocessor requirements. We also added new functionality to continue retrying query construction following errors, improving recall.

We added new functionality to the AI web app to side-by-side test different model contexts. This can be used to improve and adjust the data model to optimize query results for natural language use cases. Supports either ad hoc query or a golden test set.

Continuing down the path of making view building in the workbook clearer and faster. Now query views (SQL or UI) get a dedicated build and edit experience.

We improved the dashboard editing experience to exact match the size in production. This means all the visualization tuning will exact match the final product. Additionally, we added a UI toggle to optimize the independent mobile layout (ie won't affect the desktop layout).

We've heard lots of demand for data directly into Google Sheets without Zapier. Now using Google Cloud Run functions, we have secure data transfer into Sheets.

We're launching the new UX for conditional formatting with a first class UX on the table options, including full-row highlighting.

We're launching all the new performance data for query runtimes into production. Lots of granularity on query performance to action better experience for end users.

Several customers have requested the flexibility to build their own app nav. Now folks can DIY application navigation outside the dashboard iframe for even more custom experiences. We added always_where_filter (a forced, hidden topic filter) alongside sql_always_where. Also some additional work to improve cache hit rates.

Continuing work on branches for content for the most mature content lifecycle out there.

Continuing to make the whole app API addressable - here with connection environment building (also landed but not demoed delete folder, import content, moving content, get queries from dashboard). Now models can be deployed cross-environment with git, for multi-instance deployment or dev-staging-prod workflows.

A new admin console for moving data models between environments - best for multi-region deployments or dev-prod workflows.

We're doing some reorg on the docs to make it easier and faster to find what you need. Feedback appreciated on how we can organize for the best experience.