You can’t review a file you don’t have all of
Most of what we talk about at Agentech is file review: holding every claim file to the carrier’s own guidelines before it closes, instead of sampling a handful after the fact. That’s QA Complete, and it’s where most conversations start.
There’s a step before it that decides whether any of that works, and it’s the least interesting thing in claims operations. The file has to be complete, correctly labeled, and in the right place first.
That’s a document problem. On most teams it’s still solved by hand.
What manual triage actually costs
Documents arrive in a shared inbox. Somebody downloads the attachment, works out what it is, renames it to whatever the naming convention is this quarter, and re-uploads it into the claims management system under the right claim.
Every one of those steps is a person doing work software should do. Every one is also a place where a document gets mislabeled, filed under the wrong claim, or sits unopened for two days because the person who watches that inbox was out.
The cost shows up three ways: the hours, the errors, and the delay. The hours are the easiest to count and the smallest of the three. The other two are what actually slow a claim down.
That matters for review, because a review is only as good as the file it reads. If the photos are in the system but the general loss report is still an unread attachment in a mailbox, a reviewer, human or otherwise, will flag something as missing that isn’t missing. The file bounces to the field, somebody spends two days explaining it was documented all along, and it comes back having lost four days and gained nothing.
That isn’t a review problem. It’s an intake problem wearing a review problem’s clothes.
What Document Routing does
Document Routing is our answer to it. It ingests email and attachments automatically, uses classification and extraction to understand what each document actually is, and routes it to the right claim in the right system without anyone renaming anything.
- Multi-source ingestion. Connect several claim inboxes, or use drag and drop, or upload a ZIP.
- Automatic categorization and tagging. Each document gets a type and a confidence score, so the low-confidence ones are the only ones a person opens.
- Routing rules that match how your team already files. The logic is built around your naming conventions and your systems during setup, not a fixed template you have to live with.
- Straight into the CMS. The document lands where it belongs, named the way your team names things.
The confidence score is the part I’d point to. The goal was never to have AI decide everything. It’s to have AI decide the obvious ones, which is most of them, and hand a person the short list that needs judgment. On the routine documents, handling time goes from minutes to seconds. The exceptions still get a human.
Where routing ends and review begins
These two get conflated on almost every call, and the line between them matters.
Routing answers: what is this document, and where does it go?
Review answers: now that the file is whole, does it meet the standard?
Routing is about completeness and placement. Review is about judgment. QA Complete reads across the photos, estimate, scope, coverage, and documentation together, against the rules the carrier on that file actually wants, and returns findings that name the specific rule, show the math, and point to the evidence in the file.
That second part matters more than it sounds. An IA firm working a book of carriers is holding a different set of expectations for each one: which photos, which forms, which line items are allowed, what’s depreciable, what the onsite contact has to be given. It runs to hundreds of combinations, and today it lives in people’s heads. Routing makes sure the file is whole. Review makes sure it was built to the right rules.
You can run either one first. They don’t depend on each other, and plenty of teams start with review on closed files because it’s the fastest way to see value on work that’s already finished. But if your files are chronically incomplete by the time they reach review, fix intake first. Otherwise review spends its time flagging documents that exist and are simply in the wrong place.
Which one your team needs
It depends on where the friction actually is.
If your reviewers keep asking for documents that turn out to already exist, start with routing.
If your files are complete but you’re only reviewing a sample of them, and you suspect the rest are carrying leakage, start with file review.
Either way, the way to find out is on your own files, not in a demo. Run a batch of closed files through review and see what it comes back with. It’s short, it’s low cost, your team keeps working the way it works today, and at the end you have a findings report you can argue with.
