Haha I've built a very rudimental version of this for my personal documents. No grouping or categorizing of text areas, just PDF->OCR->Database. I do scan for letterheads, though, and save them in a separate table.

A question, if I may: How big is your userbase?