Extracted cell text · 15 estimated tokens
Quarterly sales
Region Revenue
Taipei 1234
Kaohsiung 987Format-specific converter
Upload a .xlsx file and get AI-readable Markdown tables — clean structure for financial sheets, product lists, and operating data.
Truncation, formulas, rounded displays and missing numeric tokens

Beary is waiting for a document. Choose or drop a file to begin.
Drop your Excel file here
or
Excel (.xlsx) · up to 100 MB
You can also drag and drop the file into this box.
Converting another format? Open the document converter.
Each worksheet becomes a Markdown table, with shared strings, dates, and booleans resolved correctly. Formulas are never recalculated — the cached value stored in the file is used when available, and disclosed honestly when it is not. Charts, pivot tables, and merged-cell layouts are not fully reconstructed. Very large or overly complex workbooks fail clearly with an explanation instead of getting stuck in processing. Because Markdown table syntax adds characters, the estimated token count can increase rather than decrease for some spreadsheets — the value is clean, reviewable structure, not a guaranteed reduction.
quarterly-sales.xlsx-100%
Quarterly sales
Region Revenue
Taipei 1234
Kaohsiung 987## Quarterly sales
| Region | Revenue |
| --- | --- |
| Taipei | 1234 |
| Kaohsiung | 987 |Measured synthetic fixture in the current XLSX pipeline: one heading, one table, and three rows were emitted. The signed reduction is -100% because the estimate rose from 15 to 30 tokens. The added heading, pipes, and separator row are observable output structure; this example does not claim that every workbook changes by the same amount.
Uploaded originals and conversion results are temporary. They are not kept as a permanent document library.
It is kept in temporary storage only while your job is processed, deleted immediately after processing ends, and never stored permanently. See our privacy page for the full retention policy.
A Markdown table for each worksheet, with dates and cached formula results resolved where available, plus a metadata block showing estimated tokens and the signed token change — which can show an increase for spreadsheet-heavy files.
Yes. PDF, Word (.docx), OpenDocument Text (.odt), PowerPoint (.pptx), HTML, CSV/TSV, and plain text are all supported on the document converter. This page focuses on Excel; use the document converter for anything else.
We are writing a series on getting documents into AI tools properly. Leave an address and we will send the new ones as they go up. No schedule promised, no marketing sequence, and one line unsubscribes you.