Documentation

Quick start

Open the app, then drop a CSV or Excel file on the upload zone — or click a sample dataset to explore without a file.

Supported files

• CSV, .xlsx, and .xls, up to 25 MB
• One CSV dataset, or one selected worksheet from a workbook
• Clean tabular data with a header row works best

The dashboard

Columns are profiled automatically: dates become time-series charts, categories become bar charts and filter menus, numeric columns become KPIs. Every filter drives every widget. Reset view clears filters; New file replaces the dataset and deletes the old session.

Asking questions

The chat panel answers plain-English questions with numbers computed by validated aggregate SQL. Each answer includes the executed query and its supporting values. Toggle Use all data to ignore active filters for one question.

Q: How much revenue came from the West?
A: $45,000 across 3 orders.
   Receipt: SELECT SUM("revenue") FROM dataset WHERE "region" = 'West'

Limits and privacy

• Sessions are ephemeral and expire after 30 minutes of inactivity
• Questions run one to three read-only aggregate SELECT queries
• Generated SQL is parsed, column-checked, row-capped, and time-limited before execution
• Raw rows are never sent to the AI provider — only column metadata and aggregate results