Query Builder

For the people who ride the SQL

A proper editor over every source you have connected, with autocomplete that has read your schema and a check that runs before your query does.

Acme’s Workspace

Search⌘ + K

John Doe

Revenue by Payment Method

1SELECT
2  TO_CHAR (DATE_TRUNC ('month', o.order_date), 'Mon YY') AS MONTH,
3  p.product_name,
4  SUM(oi.total_price) AS revenue
5FROM
6  ecommerce.orders o
7  JOIN ecommerce.order_items oi ON o.order_id = oi.order_id
8  JOIN ecommerce.products p ON oi.product_id = p.product_id
9WHERE
10  o.status = 'fulfilled'
11GROUP BY
12  DATE_TRUNC ('month', o.order_date),
13  p.product_name
14ORDER BY
15  DATE_TRUNC ('month', o.order_date) ASC,
16  revenue DESC;

Revenue

20M15M10M5M0

Oct 24Nov 24Dec 24Jan 25Feb 25Mar 25Apr 25May 25Jun 25Jul 25Aug 25Sep 25Oct 25Nov 25Dec 25

Give me the top 10 products sold in the last month

Identifies the maximum order_date, then aggregates quantity and revenue for fulfilled orders within the prior month, grouping by product and sorting by units sold descending.

Change +37

+ WITH

+ max_order_date AS (

+ SELECT

remove the fulfilled filter

Removed the o.status = 'fulfilled' condition and its trailing AND from the WHERE clause to include all orders regardless of status.

Change +1 -2

- o.status = 'fulfilled'

- AND o.order_date >= (

+ o.order_date >= (

pivot this

Returns top 10 products sold in the last month with product_id, product_name, total_units_sold, and total_revenue as row-level metrics grouped by product.

Change +1

+ AND o.status = 'fulfilled'

Ask here...

Ask Analysts Datasites Dashboard Automations Query Builder Agents

Trusted by 1,000+ businesses across the world

The query is easy. Everything around it isn't.

Disconnected database and platform shapes scattered across a grid

The join lives in one person's head

Two tables, one correct path between them, and the only record of it is whoever built the warehouse.

A briefcase labeled data insights locked away from a business stakeholder

Write, run, fail, repeat

Nothing tells you the column is misspelled until the query comes back empty twenty seconds later.

Mismatched abstract shapes that refuse to fit one visual style

Good queries get lost

The right version of last quarter's query is in a chat thread, a notebook, or nowhere.

An editor that has read your database

Write it in the language you'd choose

SQL for the query, Python when the answer needs more than SQL. One editor, over every source you have connected, with autocomplete that has read your actual tables and columns.

SQL and Python
Every connected source in one place
Autocomplete from your real schema

A syntax-highlighted SQL query aggregating order count, revenue and average order value from the last sixty days of shipped orders

Ask for the change, read the diff

Say what the query should also do and the assistant proposes the edit as a diff — the removed line beside the added ones — so you approve a change you can read instead of a rewrite you have to re-check.

Edits arrive as a reviewable diff
It knows the schema, so the column exists
Checked before it reaches the database

The assistant adding an average product price column, shown as a diff with one line removed and two added

The query doesn't end at the results grid

Switch the result to a chart or a KPI, pin it to a dashboard, or save it as a verified query the AI analysts reuse instead of writing their own.

Results into a chart in one step
Saved with variables for reuse
Verified queries the analysts reuse

A SQL query grouping revenue by month and plan, with its result already rendered as a grouped bar chart below the editor

What you can point it at

700+ Sources in one editor
Every database and business app in the catalogue, queried from one place.
123 You can connect yourself
No call with us: pick it, authorise it, query it.
15 Minutes to your first query
Connecting a source is the only setup there is.

Frequently asked questions

Who is the query builder for?

People who already write SQL and don't want to stop. Someone who doesn't write SQL can still read what you did, because the query and the result sit side by side.

Can I use Python?

Yes. SQL and Python both run in the same editor, against the same connected sources.

What does the AI actually do here?

It completes what you are typing using your real tables and columns, and when you ask it to change the query it proposes the edit as a diff you approve line by line rather than rewriting the query under you.

What is a verified query?

A query you have saved and marked as correct. The AI analysts reuse it instead of writing their own, which is how a team's good SQL stops getting lost.

Which sources can I query?

Anything you have connected. 123 sources connect without talking to us, out of a catalogue of 700+.

Your data has the answers

Start free trial Book a demo