01
The numbers, on one screen
Summary tiles for the figures people ask for, charts for the trend, and a table underneath for the row somebody always wants to click into. Filters and a date range that apply to all of it at once.
Use case — web app
Build the operations dashboard instead of exporting another spreadsheet
The weekly numbers exist. They are just in four places, and somebody spends Monday morning turning them into a slide nobody trusts by Wednesday. A dashboard is a query, a few charts, a date filter and permissions — small enough that it never gets a roadmap slot and important enough that its absence costs a morning a week. Vibely builds it against your own Postgres database and deploys it to a URL, so the meeting looks at one live screen instead of four stale exports.
- Built against your own Supabase Postgres schema
- Filters and a shared date range across every chart and table
- Role-based access instead of a link anyone can edit
- Deploys to a live URL — nothing for your team to install
- React and TypeScript you own, synced to GitHub whenever you want it
- Change a metric by asking, and see it on the same screen
Start with this prompt
Paste it in and edit the details to match your own. Being specific about the screens and the data is what separates a working app from a mockup.
Build an operations dashboard for a support team, against my Supabase database. Top of the page: four tiles — tickets opened, tickets closed, median first response time, and open tickets older than seven days. Below that: a line chart of opened versus closed per day, and a bar chart of tickets by category. Everything is driven by one date-range picker at the top, defaulting to the last thirty days, plus a filter by team. Underneath, a table of the currently open tickets, sorted oldest first, that I can click into. Sign-in with two roles: a lead sees their own team, an admin sees all teams.
What to build next
- Add a CSV export of whatever the filters are currently showing.
- Add a comparison against the previous period on each tile.
- Add a saved view so each lead lands on their own team by default.
Web app
What the build actually consists of
For an operations, finance or support lead whose data is already in a database or a CSV somebody exports on a schedule. If the data does not exist yet, build the tool that captures it first — an internal tool or a tracking app — and the dashboard follows from it.
02
Over your own data
Point it at your own Supabase Postgres project, or let Vibely provision one and load a CSV into it. The agent can read the live schema, so the charts are built against your actual tables and column names.
03
Scoped by role
Sign-in with roles, so a team lead sees their team and a director sees everything — rather than the current arrangement, where the numbers are in a sheet anyone with the link can edit.
What you get
- Summary tiles for the figures the meeting opens with
- Charts for the trend, with a date range that drives all of them
- A drill-down table for the rows behind a number
- Sign-in with roles, so access is scoped by team
- A live URL the meeting can open instead of an export
Describe it. Watch it build.
Paste the prompt above and react to something running instead of writing a spec. The free plan needs no card, and the code is yours from the first build.
Start buildingFAQ
Questions, answered.
Can it read our existing database?+
Yes, if it is Postgres on Supabase — link your own project and the agent can read the live schema and build against your real tables. For data elsewhere, the usual first move is a CSV import into a Supabase project so the dashboard has something to query.
Is this a BI tool?+
No, and it should not be sold as one. It is a purpose-built screen for the handful of numbers your team actually reviews, which is what most "we need BI" requests turn out to be. If you need ad-hoc exploration across a warehouse by analysts who write their own SQL, buy a BI tool.
How fresh are the numbers?+
They are read from your database when the page loads, so they are as current as the data is. There is no sync step and no extracted copy to go stale.