Documentation

AI Data Studio is a browser-based SQL tool powered by DuckDB. Everything runs locally in your browser — no data is sent to any server.

Getting Started

Upload a CSV or Parquet file using the drop zone in the sidebar.
Your file appears as a table in the schema explorer with column names and types.
Click a table name to auto-fill a SELECT query, or write your own SQL.
Press Cmd + Enter or click Run to execute.
View results in the table below, then export as CSV, JSON, or Parquet.

Features

In-Browser SQL Engine

AI Data Studio runs DuckDB entirely in your browser using WebAssembly. No server, no data leaves your machine. Your queries and data stay 100% local.

File Upload

Drag and drop CSV or Parquet files into the sidebar. Each file is registered as a table you can immediately query with SQL. Multiple files can be uploaded at once.

SQL Editor

Write SQL in a full-featured editor with syntax highlighting. Click the Run button or press Cmd/Ctrl + Enter to execute. DuckDB supports standard SQL including joins, aggregations, window functions, and more.

Schema Explorer

The sidebar shows all tables with their column names and data types. Click a table name to auto-populate a SELECT query. Tables are automatically shown after file upload.

Export Results

Export query results in three formats: CSV for spreadsheets, JSON for APIs and programming, or Parquet for efficient columnar storage. Buttons appear in the results toolbar.

Data Persistence

Uploaded files are stored in IndexedDB, while your editor text and query history are saved to localStorage — all in your browser. Refresh the page and everything is still there.

Query History

Every query you run is saved (up to 50). Open the Query History panel in the sidebar to browse previous queries and click any entry to reload it into the editor.

Dark / Light Mode

Toggle between dark and light themes using the button in the top-right corner. The editor and all UI elements adapt to your preference.

Keyboard Shortcuts

Cmd/Ctrl + EnterRun the current SQL query

SQL Examples

Select all rows

SELECT * FROM "my_table" LIMIT 100;

Count rows

SELECT COUNT(*) FROM "my_table";

Filter and sort

SELECT * FROM "my_table"
WHERE status = 'active'
ORDER BY created_at DESC;

Aggregate by group

SELECT region, COUNT(*) AS total, AVG(amount) AS avg_amount
FROM "my_table"
GROUP BY region
ORDER BY total DESC;

Join two uploaded files

SELECT a.*, b.category
FROM "orders" a
JOIN "products" b ON a.product_id = b.id;

Create a table from a query

CREATE TABLE summary AS
SELECT region, SUM(amount) AS total
FROM "sales"
GROUP BY region;

Supported File Formats

.csv.tsv.parquet

Files are parsed client-side by DuckDB. CSV files are auto-detected for delimiters, headers, and types. Parquet files preserve their schema exactly.

Datasets

AI Data Studio comes with a collection of curated datasets you can load instantly — no downloads or uploads needed. Visit the Datasets page to browse them.

Quick Load Datasets

Click the play button next to any dataset to load it directly into the SQL editor. The dataset is fetched from a public URL and queried in your browser using DuckDB's read_csv() or read_parquet() functions.

Query External Data by URL

You can also query any publicly accessible CSV or Parquet file by URL directly in the SQL editor. For example:

SELECT * FROM read_csv('https://example.com/data.csv') LIMIT 100;

SELECT * FROM read_parquet('https://example.com/data.parquet') LIMIT 100;

Available Categories

ClassicDemographicsEconomicsEntertainmentFinanceGeographyHealthReferenceTransportation

Data Persistence

All query history and files are stored using localStorage and IndexedDB in your browser. Nothing is sent to a server.

Uploaded filesIndexedDB

Editor SQL textlocalStorage

Query history (last 50)localStorage

On page load, saved files are automatically re-registered with DuckDB so your tables and schema are restored. Clearing your browser data will remove all stored files and history.

AI Assistant (Bring Your Own Key)

AI Data Studio includes an AI assistant that can generate SQL queries from natural language. This feature requires your own Anthropic API key — no key is provided by default.

Getting an API Key

Go to console.anthropic.com and create an account.
Navigate to API Keys in the dashboard.
Click Create Key and copy your new key.
Paste the key into the API Key field in the AI Data Studio sidebar.

Available Models

You can choose between different Claude models in the sidebar. Smaller models like Haiku are faster and cheaper, while larger models like Sonnet produce more accurate queries for complex questions.

Privacy: Your API key is stored only in your browser's localStorage. It is sent directly to the Anthropic API to generate SQL — it is never stored on any server. Your table schemas are shared with the AI to provide context, but your actual data is never sent.

Privacy

All processing happens in your browser. Your files and queries never leave your machine. There is no backend server, no analytics, and no data collection.