Supabase logo

Dataki for Supabase

Ask Supabase in plain English

Connect your Supabase Postgres database and turn it into dashboards. Sign in with Supabase to pick the project, add its database password, and Dataki queries your tables directly.

Connects directly Sign in with Supabase, then add the database password.

Connect Supabase Read the setup guide
app.dataki.ai / chat Supabase · public

How many people signed up each week for the last eight weeks?

Reading the schema, writing the query…
SQL 190 ms · read-only 
SELECT DATE_TRUNC('week', created_at) AS week, COUNT(*) AS signups
FROM   profiles
WHERE  created_at >= DATE_TRUNC('week', NOW()) - INTERVAL '7 weeks'
GROUP BY week ORDER BY week;
Signups by week 8 rows
Aug 3Aug 10Aug 17Aug 24Aug 31Sep 7Sep 14Sep 21 0100200300400500600

Weekly signups went from 312 to 544 in six weeks; this week has 521 so far.

01Questions

What people ask Supabase

Each one becomes a query against Supabase, run read-only, with the SQL shown beside the chart so anyone who reads SQL can check it. Anything worth keeping goes on a dashboard that refreshes on its own.

  • How many people signed up each week this month, and how many came back?
  • Which plan do our most active users pay for?
  • How many projects did each plan create in the last 7 days?
  • Which organisations added the most members this month?

02From your assistant

Ask Supabase from Claude, ChatGPT or Cursor

Dataki runs a hosted MCP server. Add it to your assistant once and it can read your Supabase schema and run read-only queries through Dataki. The assistant gets the rows back; the Supabase credentials stay encrypted in Dataki and never reach the model.

Setup for your assistant:

Claude · dataki (MCP) Supabase · public

How many new users did we get each day this week in Supabase, and which sign-up method did they use?

  1. list_sources · finds Supabase · public
  2. get_data_context · reads the tables and what the columns mean
  3. query · runs the SQL read-only, returns the rows
  4. render_ui · draws the chart from those rows, in the chat
server URL streamable http
https://dataki.ai/api/mcp

03Setting it up

Connecting Supabase

Sign in with Supabase, then add the database password. The setup guide covers it step by step.

Setup
About 10 minutes, the read-only role included.
Connects from
34.89.253.13, over IPv4. Allow it if you use network restrictions.
Encryption
TLS: Dataki offers it first, so Supabase's Enforce SSL can be on. The server certificate is not verified.
What it reads
Tables and views in the public schema, and only the rows row level security lets its role see.
Read the Supabase guide Open Dataki

The answer is still there tomorrow.

Ask in plain language. Dataki writes the SQL, runs it against your warehouse and draws the chart. Anything worth keeping becomes a dashboard with a link — not a message that scrolls away.

app.dataki.ai / chat

Which channels actually brought in revenue last quarter?

Queried bigquery · payments · 1.2 GB scanned 
SELECT channel,
       SUM(amount) AS revenue
FROM   payments
WHERE  created_at >= DATE_TRUNC(CURRENT_DATE(), QUARTER)
GROUP BY channel
ORDER BY revenue DESC

Revenue by channel

Q3 · 5 rows

Organic search
$412,900
Partner referral
$268,400
Email
$151,200
Paid social
$96,700
Direct
$61,300

One click later, that chart is a dashboard of its own:

app.dataki.ai/d/revenue-by-channel

Live, not a screenshot. It re-runs the query on a schedule, so the number your team reads on Friday is Friday's number.

  • Share the link with anyone. Viewers are free on every plan.
  • Edit the SQL by hand whenever the model gets the logic almost right.
  • Drop the same dashboard into your own product with one line of HTML.
Start free See how it works

Using Dataki with Supabase

Can I query Supabase in plain English instead of writing SQL?
Yes. Dataki reads your Supabase schema, turns a plain-English question into a query, runs it against Supabase and returns a chart. The generated query is shown with every result, so you can verify or edit it.
Is there a hosted Supabase MCP server?
Dataki is one. Its remote MCP server reaches Supabase through a read-only connection you set up once in Dataki, so nothing is installed on your machine and no connection string goes into an assistant's config. It offers tools to list sources, read the schema and its documented meaning, run read-only SQL, draw charts in the conversation and save dashboards. Every query is logged with the key that ran it.
How do I build a dashboard from Supabase data without an ETL pipeline?
Connect Supabase to Dataki with a read-only connection, ask your question, and keep the answer as a dashboard widget. Dataki queries Supabase directly at read time, so there is no pipeline to build, no warehouse to sync into and no copy of your data to keep current.
Can Claude, ChatGPT or Cursor query my Supabase data?
Yes, through Dataki's hosted MCP server at https://dataki.ai/api/mcp. Connect Supabase to Dataki, add that URL as a connector in Claude or ChatGPT (or in Cursor, VS Code or Claude Code with a Dataki key), and the assistant can read the schema and run read-only queries. It gets the rows back; the credentials stay encrypted in Dataki and are never handed to the model.