CSV, Excel, Google Sheets
Order exports, ad exports, payouts, ledgers, anything with a header row and rows under it.
Upload a spreadsheet, ask in plain English. It writes and runs real Python — every number opens to its code.
Sample data, no install, no call — or start free.
Not a recording. Both panes below are the product, computing sample data in your browser; drop your own file in and every figure recomputes on it.


A chat model estimates. The Analyst writes pandas, runs it against every row, and hands you the code with the answer.
KPIs, forecasts and anomaly flags come out of code that ran on your rows.
Every figure opens to the query that produced it. Copy it, re-run it, argue with it.
Ask for the chart, the slide or the projection; keep any of them.
Your file is parsed and computed in your browser; only what an AI step needs leaves it.
Frontier models from Anthropic, OpenAI and Google, per question, or let it route.
Set a rule on a metric; it re-runs on the next upload and emails you when it trips.
Pick the model that fits the job, from ultra-cheap and fast to deep-reasoning. Whichever you choose, every number stays computed from your data and shows its work.
No setup, no connector, no schema to define. The file you already have is the input.
CSV or Excel, straight off your desktop. It loads into a Python runtime that runs in your browser, so the rows never leave the tab.
The Analyst writes real pandas and executes it. You never write SQL, and every figure it returns opens to the code that produced it.
Export the deck, put the question on a schedule, or set a rule on any KPI. The work survives the session instead of dying in a chat log.
Anything that comes out as a spreadsheet. No OAuth to grant, no pipeline to build, nothing to wait on.
Order exports, ad exports, payouts, ledgers, anything with a header row and rows under it.
It maps your headers itself and tells you what it mapped them to, so a file nobody cleaned still computes.
The numbers are computed in your browser. Saved work is encrypted and PII-masked, kept in your account until you delete it, and never used to train a model.
One credit is one AI action. Free is free for good; annual is ten months for twelve.
Drop a spreadsheet and get answers with the math attached.
Start free