Whereas pdftotext tools have been there since forever.
I've written tools to try and extract text from docx files programmatically (not CVs but citations and footnote citations) and I seriously doubt "hiring startups" wouldn't just ask you to fill a web form.
(As a matter of fact, I'm low key job hunting and what I've seen typically is that the systems try to extract the text from my non-standard formatting latex-made pdf, and then pre-fill a form. The form is usually right but needs fixes here and there. It's a good approach.)
The thing they hate most in life is to be forced to do actual human work, especially when they are processing all those résumés and CVs.
I mean, they have to feed the machine somehow, and you’re making them do real work during that process?!?