Point this at a photo of a table, receipt, or list, and it reads the text via on-device OCR, splitting it into spreadsheet-style rows, a fast way to digitize a printed table without retyping it.
Image quality drives accuracy more than anything else. A photo taken straight-on, in good light, with sharp focus reads far more reliably than a blurry or angled shot. If the extracted text looks off in a few spots, retaking the photo usually beats fixing every error by hand.
A pile of paper receipts that need to go into an expense spreadsheet, or a printed price list that needs to become a real product database instead of getting retyped by hand.
It depends on image clarity. Sharp, well-lit, high-contrast photos of printed (not handwritten) text give the best results.
It splits each line wherever it finds several consecutive spaces or a tab, which matches how most tables and receipts are naturally spaced.
Yes, the recognized text shows up in an editable box before you export the CSV.
It's built for printed text. Handwriting recognition is much less reliable.