Golden shipped a massive week — multi-table support with AI relationship detection, a new Matrix Heatmap chart type, a completely rebuilt data cleaning wizard with streaming AI suggestions, custom workspace color palettes, Excel-like table interactions across the platform, an overhauled export system, and significant warehouse connector fixes.
Check it out:
The data cleaning wizard has been completely rebuilt. It now streams suggestions as they're generated (first results appear in ~3 seconds instead of waiting up to a minute), shows real before/after previews for every proposed change, and automatically detects the purpose of your data — so the lengthy upfront questionnaire is gone. Suggestions you don't want can be individually dismissed.
Workspace admins can now create and manage custom color palettes in Team Settings to match your brand guidelines. Palettes appear everywhere colors can be set in the app — chart color encoding, format pane, and dashboard properties. Extras include AI palette generation from a brand description, color extraction from an uploaded image, accessibility scoring with color-blindness simulation, and drag-to-reorder colors.
Exporting dashboards and stories is now significantly higher quality:
All tables in Golden now behave more like a spreadsheet: click any cell to select it (highlighted with a green border), use Ctrl+C to copy selected cells to the clipboard, and see a live status bar at the bottom showing Sum, Average, Count, Min, and Max for your selection.
A new Matrix Heatmap chart type is now available. Map a row field, column field, and value to produce a color-coded grid — great for showing patterns across two categorical dimensions (e.g. day of week × hour of day). Options include custom color scales, value normalization, cell value labels, and resizable rows/columns.
When connecting to a data warehouse (Snowflake, Databricks), you can now select multiple tables in one schema import. Golden automatically detects relationships between tables using foreign key metadata and AI inference, and displays a relationship editor so you can review and adjust before importing.