Dashboards

Dashboards are the most powerful flagship feature in Dreambase. You will learn how to quickly and easily select topics from the data dictionary to rapidly generate a production quality dashboard which is completely editable an refreshable to fetch new data on demand or on a schedule.

Dashboards are the flagship, primary production layer in Dreambase. Point Dreambase at your data (Supabase tables, connected APIs, MCP tools, or a Skill's curated context) and get a production-quality, editable, refreshable dashboard on demand or on a schedule.

Creating a dashboard

Start from the homepage chat or a Skill. The homepage composer ("One question away from clarity") accepts a prompt plus # to mention specific tables and / to mention a Skill for pre-built context. Suggested prompts are shown inline (e.g. "Generate a dashboard for /Court Utilization").
Write a short prompt describing what you want to see (e.g. "Show me monthly subscription trends and revenue breakdown by plan"). The AI assistant uses your prompt plus whatever tables/Skills you referenced to generate the dashboard.
Background generation. Dashboards generate in the background — you can close the page safely and you'll be notified when it's ready.

Correction from the previous version of this doc: dashboards were previously described as being built from a "Topics" semantic layer, created during onboarding or in a separate "Data Dictionary." In the current live app we did not find a standalone Topics/Data Dictionary section — the equivalent context now comes from # table mentions (raw schema) and Skills (curated, reusable semantic layers with their own data sources and AI-readable docs — see the new Skills doc). If "Topics" still exists as a backend concept under a different label, let us know so we can reconcile the terminology exactly.

The Analyst Agent (per-dashboard chat panel)

Every dashboard has a right-side Analyst Agent chat panel — this is new since the last version of this doc and wasn't documented before:

Attached context shows which Skills/tables are informing this dashboard.
The composer supports # for tables, / for Skills, and @ for other mentions, same syntax as the homepage.
Quick-action prompts appear inline (e.g. "Change the chart widths", "Add more filters from my columns").
A Toggle edit mode control switches between viewing and editing the layout.
Refresh schedule options let you set how often the dashboard's underlying data refreshes.
Notification settings control when you're pinged about this dashboard (e.g. on refresh, on errors).
If a metric renders empty, a "Why this metric is empty" affordance explains the gap directly in the UI.

Editing your dashboard

Drag-and-drop edit mode: resize/rearrange components, add/remove metrics, adjust styling.
AI assistant refinement: modify the dashboard using natural language in the Analyst Agent panel, using the same / and @ mention syntax.
Example prompts: "Add a chart showing daily signups for the last 30 days", "Change the revenue chart to a bar chart grouped by month."

Dashboard components

Metrics — single-value KPIs with optional trend indicators.
Charts — line, bar, pie, and area charts.
Lists — flexible lists with sorting and filtering.
Data tables — full-screen tables with pagination, search, and export.
Links — references to other dashboards or external resources.
Filters — dashboards can expose filter controls tied to your data (e.g. "Email Verified: Y/N", "Marketing Opt-In: Y/N") that viewers can toggle.

Data refresh and caching

Dashboard queries save results as datasets in Supabase Storage (the Transforms layer — see its own doc). Subsequent loads read from this cache to lower cost and reduce load on your production database.
Manual refresh anytime, or scheduled refresh (hourly, daily, weekly, etc.) via the dashboard's refresh schedule options.

Managing dashboards

Pinning: pinned dashboards stay at the top of the homepage sidebar for fast access — confirmed live, with a dedicated "Pinned" section showing refresh timestamps and read/unread state per dashboard.
Grid and list views with search across names/descriptions and filters/sorting.
Free-plan workspaces will see an "Upgrade to add more dashboards" prompt once they hit their dashboard limit.

Best practices

Reference specific tables or a Skill (#//) rather than writing free-form prompts with no context — it produces more accurate first drafts.
Combine related tables/Skills to surface connected business metrics in one view.
Set a refresh schedule that matches how often your underlying data actually changes.
Iterate with the Analyst Agent rather than rebuilding from scratch.
Pin your most business-critical dashboards.