It seems to meant for end-users like journalists processing files individually like e-mail attachments.

It doesn't seem to be meant for usage at scale -- it's not for general-purpose conversion, as the resulting files are huge, will have OCR errors, etc.