Dashboard Canvas
Drag, drop, resize, and filter tiles into live dashboards.
The Dashboard Canvas turns the charts and tables Databasin One makes into tiles, lets you arrange and filter them in a grid, and publishes the whole thing to your team's Gallery. It's the fastest path from "ask a question" to "share an answer."
How it starts
Most chart and table answers in Databasin One have an Add to Dashboard button. Click it and the answer becomes a tile; the button flips to On Dashboard so you know it's there. The canvas opens as a modal over the chat.
You don't have to start from chat, though. The canvas also has an Add Visual button that opens a small creator — describe what you want in plain language or paste SQL, and it builds a tile in place.
Working with tiles
Each tile is one chart or table. In the tile header and corner you can:
- Drag to reorder — pick up any tile and drop it elsewhere in the grid.
- Cycle its size — a button in the header steps through preset sizes: medium → large → wide → small.
- Resize it freely — drag the corner handle to set the width (1–4 columns) and height (200–1200px) directly.
- Rename it — double-click the title.
- Remove it — the X in the header.
Above the tiles, two more actions apply to the whole dashboard: Export PDF and Publish.
Tiles are the exact chart
When you add a chart, the canvas stores the exact ECharts option object that rendered in chat — no re-mapping, no re-translation. That's why a tile looks identical to the chat version: it is the chat version.
This is deliberate. The older dashboard path (ReportChart / ReportDashboard) re-mapped charts through an intermediate layer, and subtle things broke — colors, series order, axis formats. The canvas sidesteps all of that.
Snapshot on the canvas, live in the Gallery
On the canvas, each tile keeps a local snapshot of its data so it renders instantly without re-querying. Canvas state is saved per browser — it's your working draft, not yet shared.
Publishing changes that. When you publish, the canvas stores the SQL behind each tile and strips the embedded data out. In the Gallery, tiles re-run that SQL under the viewer's permissions — so the dashboard stays fresh, and each viewer only ever sees the rows their own access allows.
Because published tiles re-run their SQL as whoever is looking, a published dashboard inherits row-level access — it never ships a frozen copy of the data. Two people can open the same dashboard and correctly see different numbers.
Filters and cross-links
A dashboard can carry its own filters. When your tiles share filterable columns, a filter bar appears above them:
- Categorical filters — multi-select the values you want to keep.
- Numeric filters — set a min and a max.
By default a filter only touches tiles that actually have that column. Cross-links stitch tiles together: pick two or more columns that mean the same thing across different tiles (say customer in one and account in another), and filtering one filters them all. Filters and cross-links ride along to the Gallery, where viewers can adjust them too.
Publishing to the Gallery
Click Publish and you get one small dialog asking for two things:
- Dashboard Name
- Description (optional)
That's it — there's no visibility or tag step here. Publishing saves the dashboard to Databasin's reports service (not just your browser) and lands it in the project's Gallery. See Publishing to Gallery for who can see it and how sharing works.
There's no "duplicate to canvas" step for editing. To update a published dashboard, change it on the canvas and publish again with the same name — that overwrites the existing entry and bumps its version.
Export as PDF
Export PDF uses the same document renderer as agent reports. It converts each chart to a base64 image, applies a print theme (white background, dark text), turns off animations, and adds a cover page with the dashboard title and the date. The output is a clean PDF for decks and attachments.