Skip to main content

v0.11.2

Experiments become easier to find and follow, with tags, Browse and search, each metric charted over time, cancellable evaluations and a locked definition once a test runs. Warehouse metric cards are fully supported on dashboards, and insight and dashboard filters now offer every segment's options.

Added​

  • Find experiments. Tag experiments and find them with filters, Browse and search.
  • Experiment metrics over time. See how each experiment metric develops day by day.
  • Experiment verdicts. Cancel a running evaluation, read results grouped into verdict categories, and rely on a definition that locks once the experiment starts.
  • Warehouse metrics on dashboards. Add warehouse metric cards to dashboards.
  • Hour-over-month comparisons. Ask the Mitzu Agent to compare an hourly chart with the month or year before.
  • JSON-string arrays. Columns that store arrays as JSON strings are read as arrays.
  • Weekly active users. The Statistics page shows weekly active users and counts AI prompts and scheduled refreshes as activity.

Changed​

  • Insight and dashboard filters offer the options of every segment and card, reach the same steps as breakdowns, and apply to dashboard cards all or nothing.
  • Statistics cards load in parallel.
  • Sidebar entries show their titles on hover when the sidebar is collapsed.
  • Dashboard card status and warning icons use a lighter orange.
  • The Mitzu Agent always asks for confirmation before overwriting or deleting, and the setup agent explains why a table could not be read.

Fixed​

  • Calculations over a broken-down funnel line up by breakdown value.
  • Experiment conversion over time works on BigQuery.
  • Scheduled agents still find their dashboard after it is renamed, and a prompt sent from the home page no longer shows up as an empty answer.