Something that seems like is should be a common use case for "AI" (computer vision, machine learning varieties)
Input: a photo or scan of a document
Output: marked-up text for the same document (word processor format or HTML)
What are the best tools for that?