The file that reads itself

Every loan generates a pile of paper that somebody has to turn into data. Mortgage statements, payoff statements, homeowners declarations both original and updated, drivers licences, notes, title commitments and CDs, tax certificates, FHA refi credit queries, lock confirmations.

Eleven document types, on every file, read by a person and typed into a form by the same person. Then checked by a second person, because the first one was typing quickly.

What we built instead

Dromi reads those documents into structured loan data. Uploads from the borrower portal sort themselves into the right folder, per document, per borrower, without anyone naming a file.

Then it stops and waits. Every extracted value shows up as a pre-filled field, not a decided fact. One click marks it reviewed and records who approved it and when.

Why the approval step is not optional

It would be faster to let the extraction flow straight through. It would also be the wrong product. Automation on a mortgage file has to be reviewable, because when a number is wrong the question is never just what went wrong, it is who signed off and when.

So nothing unverified ever drives an action. That is the design, and it is not going to change as the models get better.