Here's everything new and improved in Golden Analytics this week.
Spaces has been significantly upgraded. You can now create two new kinds of spaces: Public spaces (discoverable and accessible to all teammates) and Collaboration spaces (membership-based, with access controlled by invitation). When you create a new Collaboration space, the Members panel opens automatically so you can start inviting your team right away without extra clicks. Organizations can also now define account-level user groups, making it easy to manage permissions for a team or department across multiple spaces at once.
The AI Chat panel can now control nearly every visual formatting option directly. You can ask it to: show or hide the legend and set its position; control data labels (which points to label, font, size, color, alignment); show or hide axes, axis titles, and gridlines; set chart title typography (font, size, color, bold/italic); style axis tick labels; adjust chart opacity and background color; set per-category color assignments ("make Sales blue and Profit red"); control bubble and mark sizing; set bar corner radius, line width and style, and area fill opacity; control table options like totals, zebra striping, and compact mode; and more. Setting a chart background color via chat now also applies on the Analyze canvas, not just on dashboards.
You can now drag a CSV, Excel, or other supported file directly onto the Home page, the Create Analysis → Connect to Data screen, or the ERD view to import it — no need to click through the file picker. Dropping a file on Home or Connect to Data creates a new analysis from that file. Dropping onto the ERD view appends the file as a new table in your existing multi-table dataset. Multi-sheet Excel and Google Sheets files continue to use the sheet picker on drop.
Press ⌘K in the Analyze view and search "View SQL" to see the exact query Golden is sending to your database for the current visualization. The dialog shows syntax-highlighted SQL, labels the dialect correctly (DuckDB, Snowflake, BigQuery, Redshift, ClickHouse, etc.), and includes a copy-to-clipboard button. The SQL is captured across all chart types and query paths.
The treemap chart type has been completely rebuilt. Highlights: a dedicated Label shelf lets you choose exactly which fields appear as cell labels (previously two lines were auto-generated). Tooltips are now sticky and filter actions from sticky tooltips are undoable. Per-cell color applies when a Color field is set; single-color cells when not. Multi-select with Cmd+click works correctly. Min/max label size controls are available in the Format pane. All controls — Label shelf, color encoding, and format options — are fully available on dashboards and in Golden Docs.
The retention chart now matches what you'd expect from dedicated product analytics tools. New options: N-period retention (did the user return at exactly period N?) vs Classic retention (did they return at any point up to period N?), configurable time granularity (day/week/month/quarter, or auto), count vs percentage display mode, an Average row showing weighted retention across visible cohorts, and a minimum cohort size filter to exclude small samples from the analysis. Retention tiles on the dashboard canvas now show the standard chart hover toolbar (type, format, AI explain).
KPI and metric cards on AI-generated dashboards now automatically format numbers compactly. $640,817,865,801.54 displays as $641B. Golden infers the right format from the field name: revenue-related fields get currency formatting, rate/percentage fields display as %, and all large numbers use compact notation. Existing custom formats are preserved.
In workspace-scoped analyses, each dashboard sheet can now be toggled between Draft and Published directly from the Analysis sidebar or the dashboard toolbar. A colored pill badge shows the current state at a glance. Only Published dashboards are visible to collaborators browsing the space.