AI and OCR tools

What you are trading, and why

Every other tool on this site runs on your device precisely because it can. These cannot. Reading handwriting, a printed table or a screenshot of an interface needs a large vision model, and that model does not fit in a browser tab. So the trade is explicit: your content is transmitted to an external provider, and each page says so before you submit anything.

What we can control, we do. The handlers for these routes never log the uploaded image, the extracted text or the generated result. A tool whose provider is not configured fails closed — it returns an honest error rather than an invented answer, and any credits held for it are refunded rather than spent on nothing.

No accuracy figure is claimed for any of these tools, because none has been independently measured. That is why every result is editable and shown next to your source: checking it is part of the job, not a sign something went wrong.

Getting a better result

Photograph straight on in even light, crop tightly to the content you want read, and set the correct language where the tool asks for one. A skewed or shadowed image costs far more accuracy than untidy handwriting does, and a page read as the wrong script produces nonsense rather than a near miss. For a very dense page, two photographs usually read better than one crowded one.

Language configurations

Handwriting recognition accepts 141 language settings and the translator accepts 109 locales. These are settings on two tools, not additional tools, and no recognition or translation quality figure is claimed for any of them.

Browse the 141 handwriting languages