I am fascinated by the arena - and will throw my two cents in here for anyone who stumbles across this thread. I think AI/ML offers the potential to solve one of the most intractable problems along these lines. Not just some customer records with a few fields, but think about reconciling dozens or hundreds of hard drives, with millions of files, used by teams of hundreds of users sharing files back and forth (creating unnecessary duplicates in the process). How could we use AI to turn those millions of files into thoughtfully tagged/organized, indexed, culled and cleaned up into smaller groups, reducing storage, increasing accessibility and discoverability?

Current tools for de-duping focus on file sizes, modification dates, and some binary parity checking. What if AI looking inside these files could do image-image, content-content comparisons and assess which were newer, older, better, etc.

If such work and/or research exists, please point me in the right direction. If not and you have the skillset or interest in this area, I think there is a fundable startup in this, please get in touch.