I have an existing corpus of about 25 million words across some 6,000 files (much of it dealing with regulations and processes) which is carefully categorized in DevonTHINK. Most of them are PDFs and DT Pro has powerful OCR and rapid search capabilities which make finding information in those documents nearly trivial.
I use Obsidian as a Zettelkasten and occasionally pull an item from my DT database for extraction into ZK notes. When I do that I reference the x-callback-url for the DT file in the note.
Basically files in DT can (when needed) become sources in my existing source processing workflow.