Build a dashboard

Build a dashboard or internal tool that reads data and shows it as tables, charts, and filters. You describe what to display and PlutoAI builds the views and the queries behind them.

When to use it

Use this for internal tools and reporting: an admin panel, a metrics dashboard, a customer or inventory browser, an ops tool your team uses. If your product is customer-facing with per-user accounts, see Build your first project.

What you need

• An account at plutoai.one.
• A connected data source. For a database, connect Supabase or Neon. If your data already lives in a spreadsheet-style store, connect Airtable. See Add a database.

Connect the data source first so the agent builds against your real tables and fields, rather than guessing at a shape.

Steps

1. 1
   Connect your data source

   Connect Supabase, Neon, or Airtable from the integrations catalog.

2. 2
   Describe the views

   Say what to show and how to slice it: the metrics, the tables, the filters.

3. 3
   Review the numbers

   Check the values in the preview against what you expect.

4. 4
   Refine the filters and layout

   Add date ranges, search, sorting, and grouping one change at a time.

Example prompt

Build an internal dashboard that reads from the orders and customers
tables in the connected database. Home page shows four KPI cards: total
revenue, orders this month, new customers this month, and average order
value. Below that, a chart of revenue by week and a searchable table of
recent orders with columns for date, customer, total, and status.

Realistic follow-ups:

Add a date-range filter at the top that applies to the KPI cards and the
chart.

Make the orders table sortable by any column and add a status filter
(paid, pending, refunded).

Add a customers page listing each customer with their total spend and
order count, sorted by spend.

Get the numbers right

A dashboard is only as good as the numbers behind it. When you review:

• Spot-check a couple of values by hand against your source data.
• If a figure looks off, name the metric and what you expected, and ask the agent to show how it is calculated.
• Confirm the date ranges and filters mean what you think, for example whether "this month" is the calendar month or the last thirty days.

Expected result

A dashboard that shows live values from your data source, with filters and tables you can drill into.

Common problems

• A number looks wrong. Tell the agent which metric and what you expected; ask it to show how the value is calculated.
• The table is slow with lots of rows. Ask for pagination or a tighter default filter.
• You connected Airtable but want relational queries. Airtable works record by record; for heavy filtering and joins, a database like Supabase or Neon fits better. See Add a database.

Next steps