[kordlessagain]

Yup. Having worked on this for a while it’s best to extract the images of the PDF, then send them to Google Vision for extraction.

I have it working with 600 page documents.