A picture of a table — a phone photo, a screenshot, a scan — has no data structure at all. AI OCR reads the rows and columns from the image and rebuilds them as real cells, so you can stop retyping receipts, statements, and reports by hand.
Do it with AI insteadUpload the image
Drop a PNG, JPG, or screenshot into ExcelGPT and ask it to turn the table into a spreadsheet.
Let AI read the grid
The OCR detects rows, columns, and headers — even from a slightly skewed photo.
Verify numbers and headers
Spot-check totals and any column where a 0/O or 1/l could be misread.
Export to .xlsx
Save the rebuilt table and continue with formulas, sorting, or charts.
Yes — AI OCR works on photos, screenshots, and scans. Clearer, straight-on images give the most accurate extraction.
It's accurate on clean images, but always verify totals and easily-confused characters (0/O, 1/l) before relying on the data.
Yes. Data ▸ From Picture imports a table from an image file or the clipboard, and you review the result before inserting it.
Convert each image separately, then stack the results in one sheet and remove the repeated header rows.
OCR often keeps currency symbols or spaces with the digits. Remove them with SUBSTITUTE and convert the result with VALUE.