A leap forward for the AI experience:

• A huge step forward in enriching Omni models for AI. We now have field specific AI context, and helpers for value generation. The net is huge step forwards in AI quality across any use case. Additionally, we added workflow to add query examples to the model AI context for ground truth examples.

• Arielle shows the potential of the AI API for building custom natural language apps. Shoot us a note if you want the app code.

• We bulked up our docs for training Blobby. Will continue adding here, but lots of lessons to share on improving the model to invite better quality for AI use cases.

• We're now passing the model AI context to more places in the app, meaning vis-specific context can be more tactical.

Power tools for admins:

• A full UI for managing database load. We've layered in new tools for per-database and per-user throttling, making sure no one person can overwhelm the database.

• Lots of work bringing the full content editing experience to branches - now content edits can be synced end to end with dbt changes and the data model for fully robust edits to your most important dashboards.

• We added more metadata for schema refreshes to understand where time is spent or any associated errors.

• A follow up to our demo from last week. A bit more debug info on the dbt metadata Omni has access to.

• We're piping detailed query performance data back the logs - now performance can be traced from dashboards to individual queries, with full timing metadata up and down.

And plenty of front-end

• We're opening the beta for editing visualizations directly from the dashboard as we put the finishing touches on the full experience. Now fast edits like KPI layouts, dashboard widths, or really anything vis related.

• We extended date fill to more complex filter options, including compound filters (ie relative to fixed date periods).

• A beautiful new card visualization for displaying logos alongside a checklist UI.

• A few new visualizations for presenting word cloud style data

• Some new affordance in the visualization for null data, giving more optionality into displaying sparse data.