Dashboards & Charts

Dashboards combine multiple insights - charts, data tables, and cards - into one layout over your datasets, with filters and a choice of private, shared, public, or embedded access. Every widget is backed by a saved query, so refreshing a dashboard reloads its datasets instead of serving a stale export.

Datasets: what dashboards are built on

Charts and dashboards render from datasets - named, saved queries against a datasource. Create one with Save as dashboard dataset in the SQL Editor or Save as Dataset on a Colrows AI table answer. Open Datasets from the left menu to edit a dataset's SQL, run it with Get data, export the result as CSV, and set its Cache policy:

  • Manual - refreshes only when you ask.
  • On access when stale - refreshes when it is opened after the refresh interval has passed.
  • Scheduled interval - refreshes every Refresh interval (minutes).

Remove after idle (minutes) drops a cached result nobody has used for that long. With Continue serving the previous cached data when refresh preparation fails checked, a failed refresh keeps the last good result in place. The panel shows the cache status, last refresh, next refresh, and last access.

Dashboard datasets need an unrestricted datasource.

Widgets generated from a dataset read from the dataset's shared cache, which is loaded once for every viewer. Shared caches and public widgets are not available on datasources with Restricted Access on; a widget on such a dataset fails with "Restricted datasources cannot use shared dataset caches or public widget loaders". Use a datasource with Restricted Access off for dashboard datasets.

Charts

Charts, data tables, and single-value cards are generated from a dataset by describing what you want. In a dashboard's edit mode, choose Add, then Insights with AI under Create from Dataset. Select a dataset, write a prompt such as "monthly sales by region as a column chart", and click Generate. If the prompt is ambiguous, Colrows asks for clarification before it builds the view. Supported chart types: line, bar, column, area, pie, and spline. Charts support aggregations and histograms (date and range), and stay responsive across screen sizes.

Dashboards

Open Dashboards from the left menu to see your dashboards and those shared with you. Click Edit to change one, then Save or Cancel.

  • Compose - Add a saved insight, or generate a new chart, table, or card from a dataset. Drag and resize widgets in edit mode.
  • Refine a view - on a generated widget, use Update this view to describe a change. Undo and redo step through the changes.
  • Filter - add dashboard filters in edit mode; viewers apply them across every widget. Selecting a value on a chart axis filters the other widgets.
  • Inspect - View SQL shows the SQL behind a widget; the info popover shows its dataset and description.
  • Refresh - Refresh reloads the dashboard's datasets and re-renders every widget.

Share & embed

Share a dashboard from its menu in the Dashboards panel. Every dashboard has three visibility levels:

  • Private - only you.
  • Shared - specific Users or User groups; the viewer's access to the dashboard is checked on every render.
  • Public link - turn on Public Share and anyone with the Share Link can open it on a public page.

To put a dashboard or assistant inside your own application, see Embedded Assistants & Dashboards for the iframe and JavaScript-loader options and the token model.

Know what a dashboard shows.

Widgets generated from a dataset render from the dataset's shared cache, so every viewer of a shared or public dashboard sees the same rows. That is why dashboard datasets must use a datasource with Restricted Access off. A saved insight added to a shared dashboard runs under each viewer's own data access and masking policies, as Colrows AI and the SQL Editor do.