for a long time I've dealt with 'documents'
in lots of forms, where I want two things.
The 'frame' (format / outline etc) and
the text. OCR is a half way house for this.
https://learn.deeplearning.ai/courses/document-ai-from-ocr-to-agentic-doc-extraction/lesson/60su3505/introduction

Adds AI into the mix, providing the frame
around the mixed text / images etc.

Others in this field may find this intro interesting.

regards


-- 
Dave Pawson
XSLT XSL-FO FAQ.
Docbook FAQ.