Use cases
Build dashboards with AI-assisted engineering
One live view of the numbers your team argues about — built on your own data, with access rules and alerts, in code you can read and export.
Ciao is an AI-assisted engineering platform for building operational dashboards as real applications — live views over Postgres, warehouses and APIs with filters, drill-down, role-based visibility and alerts. Unlike BI tools priced per viewer, a Ciao dashboard is standard React, TypeScript and Supabase code you own, with metric definitions versioned in code and every change tested by automated QA before publish.
Published 2026-07-03 · Last updated 2026-07-03
One version of the numbers
A dashboard is a live view of the numbers a team runs on: pipeline this quarter, tickets breaching SLA, trucks on the road, units off the line. The point is not the charts — it is that everyone stops arguing about whose spreadsheet is right and starts arguing about what to do.
Most dashboards fail for one of three reasons. The metric was never defined precisely, so two teams compute it differently. The data is stale, so nobody trusts it. Or the tool prices per viewer, so the one screen that should be everywhere sits behind a license wall and gets screenshotted into slide decks instead.
Ciao builds dashboards as real applications. Metric definitions live in readable, versioned code — reviewable when someone disputes a number. Data comes straight from your Postgres, warehouse or APIs. Access is role-based, not license-based. And because it is an application, a dashboard can act: alert a channel, open a task, kick off a workflow when a threshold trips.
What a dashboard actually requires
- Data source connections — Postgres, Supabase, warehouse tables and REST APIs — queried directly, not exported into a second copy that drifts.
- Agreed metric definitions — Each metric defined once, in code, with its filters and edge cases explicit — the definition is the documentation.
- A refresh strategy — Live queries for operational counts, cached aggregates for heavy history — chosen per panel, not one global compromise.
- Filters and drill-down — From the headline number to the underlying rows in two clicks, because the first question is always "which ones?"
- Role-based visibility — Executives see company-wide, managers see their team, external partners see only their slice — enforced in the backend.
- Alert thresholds — Slack or email when a metric crosses a line, so the dashboard works even when nobody is looking at it.
- Export and embedding — CSV export for the analysts, and embedding into the portals and tools where people already work.
- A layout that survives a wall screen and a phone — The ops floor TV and the founder's commute both count.
How a dashboard build runs on Ciao
1. Describe the decisions
Not "a sales dashboard" but "daily pipeline coverage by region, SLA breaches by queue, and who owns each breach" — dashboards serve decisions.
2. Connect the sources
Point Ciao at Postgres, warehouse tables or APIs; the full-stack console shows every query the dashboard runs.
3. Pin the definitions
Each metric lands as explicit, reviewable code. When someone disputes a number, you read the definition instead of reverse-engineering a chart.
4. Build panels and drill-downs
Charts, tables and detail views refined with inspect-to-prompt while the live preview updates beside the chat.
5. Set access and alerts
Role tiers decide who sees which panels; thresholds wire into Slack and email.
6. Test and publish
QA smoke gates confirm panels render and queries return before every publish; production checks run after.
7. Iterate weekly
New metric, new filter, new panel — described in plain language, shipped through the same loop, with rollback if a change misleads.
Security and governance checklist
- ✓ Role-based visibility enforced in the backend, not hidden panels
- ✓ Read-only database credentials scoped to what the dashboard needs
- ✓ Metric definitions versioned in code with review on changes
- ✓ SSO sign-in so dashboard access follows your identity provider
- ✓ Append-only audit trail across prompts, merges and deploys
- ✓ QA smoke gates before publish; production checks after publish
- ✓ Export controls on panels containing sensitive rows
- ✓ Full code export — the dashboard is yours, including its queries
Dashboard variations teams build
Revenue operations dashboard
Pipeline coverage, stage conversion and forecast against target, drilling to the deals behind every number.
Support SLA dashboard
Queue depth, first-response and breach risk by team, with alerts before a breach instead of a report after.
Logistics fleet dashboard
Vehicles, routes, delays and exceptions on one screen, fed by telematics and order data.
Production line dashboard
Output, downtime and scrap by line and shift, visible on the floor where it changes behavior.
Marketing performance dashboard
Spend, pipeline contribution and campaign performance joined across ad platforms and the CRM.
Executive KPI dashboard
The dozen numbers leadership actually tracks, each drillable to the operating detail behind it.
Dashboard requirements, covered
| Requirement | How Ciao covers it |
|---|---|
| Live company data | Direct connections to Postgres, Supabase, warehouses and REST APIs |
| Numbers people trust | Metric definitions in versioned, reviewable code |
| Everyone can see it | Role-based access instead of per-viewer licensing |
| Act on thresholds | Alerts to Slack and email; workflows triggered from the same app |
| No broken Monday dashboard | QA smoke gates before publish, production checks after |
| Embedding | Standard React components, embeddable in your portals and tools |
| Ownership | 100% code ownership — export queries and UI to your repo at any time |
Frequently asked questions
Can Ciao dashboards query our data warehouse?
Yes. Dashboards connect to Postgres, Supabase, warehouse tables and REST APIs, and the full-stack console shows exactly which queries run. Heavy aggregates can be cached on a schedule while operational counts stay live.
How is this different from a BI tool?
Ciao is built for teams that want the dashboard to be part of a real application — with actions, alerts, role-based access and no per-viewer seat — rather than a separate analytics product. If your need is ad-hoc exploration by analysts, a BI tool still earns its place; the two can share the same warehouse.
How do we stop two teams computing the same metric differently?
Every metric's definition lives in the dashboard's code — explicit, versioned and reviewable. Changing a definition is a visible change that goes through review and QA, not a silent edit inside a chart builder.
Can different roles see different numbers?
Yes. Visibility is enforced in the backend by role — executives, managers, teams and external partners each get their slice, and access-control probes confirm the boundaries against the live app.
How do we start?
Self-serve: describe your first dashboard, connect a data source and publish with credits. Teams standardizing dashboards across departments can talk to sales when it becomes a program.