Question 1

What is Datacell?

Accepted Answer

Datacell is a programmable table: a spreadsheet grid where a column can be a formula, a Python expression, or an AI prompt. Unlike a conventional spreadsheet app, the same table can be computed by a spreadsheet formula engine, by Python running locally in the browser, and by a per-row AI model call — and the whole table can be read and written by an AI agent over MCP or REST.

Question 2

Can I run Python in a spreadsheet cell?

Accepted Answer

Yes. Write =PY("…") in a cell and it executes Python through Pyodide (a WebAssembly build of CPython) inside your browser — no server, no install. The cell can access other cells by reference, the current row, and the whole sheet as a pandas DataFrame, so expressions like =PY("df.groupby('region')['revenue'].sum().max()") work directly on your data.

Question 3

What is an AI column?

Accepted Answer

An AI column is a column whose values are generated by a language model rather than typed. You write a prompt template with {ColumnName} placeholders, choose which other columns the model should read as context, and Datacell makes one model call per row. The results are written back into the grid as ordinary cell values, so they can be referenced by formulas, sorted, filtered, exported, and edited like any other data.

Question 4

Can an AI agent use Datacell?

Accepted Answer

Yes. Datacell exposes an MCP (Model Context Protocol) server at /api/mcp with five tools — list_workbooks, get_workbook, get_sheet, update_cell, add_row — plus CSV and JSON resource templates. Every cloud-saved workbook also gets an auto-generated REST API with CRUD endpoints per sheet, and per-sheet webhooks for row.created, row.updated, and row.deleted events.

Question 5

Is my data uploaded to a server?

Accepted Answer

Not by default. Import, parsing, formula evaluation, Python execution, charting, and automations all run in your browser; workbooks persist locally in sessionStorage with an IndexedDB mirror. Cloud save to Neon Postgres is an explicit opt-in for authenticated users, and AI requests are the only path that leaves the device unless you enable cloud save.

Question 6

Do I need to install anything?

Accepted Answer

No. Datacell is a web application — open the URL, drop in a CSV or Excel file, and start working. There is no install, no local server, and no account required to use it locally.

Question 7

Does it still work like a normal spreadsheet?

Accepted Answer

Yes. Standard A1-style formulas with cell references, ranges, cross-sheet references, named ranges, circular-reference detection, and automatic recalculation through dependency chains are all supported, alongside number formats, conditional formatting, charts, pivot tables, and find and replace. The programmable column types are additive, not a replacement.

Question 8

What file formats can I import and export?

Accepted Answer

Import: CSV, JSON, Excel (XLSX and XLS), and ODS, via drag-and-drop or the file picker. Export: Excel (.xlsx), CSV, JSON, and PDF. Large CSV files are parsed in a Web Worker so the interface stays responsive.

Question 9

Can I share a table and collaborate in real time?

Accepted Answer

Yes. Cloud-saved workbooks sync in real time through a Yjs CRDT over a WebSocket, with a presence bar showing connected collaborators. You can also issue signed share links, grant per-collaborator view or edit permissions, and hide specific columns or filter rows per person.

Question 10

Is Datacell free?

Accepted Answer

Yes. Datacell is free to use, including local workbooks, cloud save, and real-time collaboration for authenticated users. No credit card is required.