OCR now shows detailed step‑by‑step progress updates (deskewing, text recognition, optional AI‑based correction), so you can see exactly what stage the job is in.
Improvements
Upload limits are now enforced based on your sign‑in status, giving anonymous users a free‑tier limit and applying the appropriate tier for logged‑in users.
AI‑based text correction runs earlier in the OCR workflow, boosting the accuracy of the corrected text.
Deskewing is automatically applied when enabled, improving image alignment before OCR.
OCR failures are handled more gracefully, preventing the whole operation from crashing.
PDFs no longer include an embedded searchable‑text layer; text extraction now relies solely on OCR, which may affect how searchable the resulting documents are.