← Back to App

Run SQL Without Installing a Database

To run one query against one file, the usual route is: install a database, create a schema, write a loader, fix the encoding, then fix the date format. There is a shorter one — open a page, add the file, ask the question.

The Ritual You Are Skipping

Route	What it costs before the first row
PostgreSQL or MySQL	Install, service, user, database, CREATE TABLE with every column typed by hand, then COPY or LOAD DATA — and a round of fixing encodings and date formats.
SQLite	Much lighter, and still a download, a command line and .import with its own quirks about types.
DuckDB	Genuinely excellent for this. Still a binary you have to be allowed to install.
Python and pandas	An environment, a package or two, and you are writing dataframe code rather than SQL.
A managed cloud warehouse	An account, a bill, and your file uploaded to someone else's infrastructure.
A browser tab	Nothing. The page is the software.

On a managed work laptop, "install" is not a step, it is a ticket. That is the real reason this exists.

Thirty Seconds, Start to Finish

1. Open the app.
2. Drag your file on. staff.csv becomes table S. Column names come from the header row; types are inferred.
3. Ask for what you want. average salary by department gives:

   SELECT department, AVG(salary) AS avg_salary
   FROM S
   GROUP BY department

4. Run it with Ctrl+Enter.

Or count staff by city:

SELECT city, COUNT(*) AS count_staff
FROM S
GROUP BY city

No schema written, no loader, no type declarations. When you are done you close the tab — and there is no container still running and nothing listening on port 5432.

What SQL Is Supported

Supported	Not supported, on purpose
SELECT with expressions, aliases and DISTINCT	INSERT, UPDATE, DELETE, DROP, CREATE, MERGE
WHERE, GROUP BY, HAVING, ORDER BY, LIMIT	Stored procedures, triggers, transactions
INNER, LEFT, RIGHT and FULL JOIN, across up to 26 files	Writing anything back to your files
UNION and UNION ALL
COUNT, SUM, AVG, MIN, MAX, and the usual string, date and maths functions
CASE WHEN, BETWEEN, IN, LIKE, IS NULL

The read-only restriction is enforced, not merely documented: anything that is not a single SELECT is refused before it reaches the engine. Your source files are never modified — they are read, and the result goes out through Export as a new file.

Good for Learning, Too

The gap between "I would like to learn SQL" and "I have a database with data in it" is where most people stop. This removes that gap: bring any CSV you already have — an export from your bank, a spreadsheet from work, a public dataset — and it is a queryable table in seconds.

The plain-English box helps here in a way an AI assistant does not: it writes the SQL in front of you, so you can read what your own sentence turned into, then edit it and see what changes. See the guide.

Reasonable Questions

Where does the query actually run?

In the browser tab, on your machine, in a SQL engine compiled to JavaScript. There is no server involved — which is also why nothing is uploaded. You can check that.

Is it fast enough to be useful?

For files up to 50 MB and a million rows, yes, comfortably. Past that the right answer is DuckDB or a warehouse, and that is a straight recommendation rather than a grudging one.

Do I need admin rights?

No. It is a web page. If your browser opens it, you can use it.

Can I save my queries?

Copy them out of the editor into a file. There is no account, so there is nowhere for the tool to store them — which is the same reason there is nothing of yours to leak.

Is the SQL dialect standard?

Close enough that ordinary queries port straight in either direction. It is not any one vendor's dialect, so very specific syntax may differ.

Does it work with no connection?

Once the page has loaded, yes, entirely. Here is that demonstrated.

Related Guides

• Plain English to SQL, with no AI
• Querying 26 files in one statement
• Running SQL with the network disconnected
• Full How To Use guide