What we touch, and what stays with your coordinator.
Every job is a bundle of tasks. The US Department of Labor publishes that bundle for each
occupation in the O*NET taxonomy. Rather than describe our scope in our own vocabulary, this
page takes the official task list for the roles that run a prior-authorization desk and marks
each task: staged by software, assisted, or left entirely with the person. It is a boundary
statement, not a capability claim.
Occupation
Staged
Software prepares it. The coordinator reviews and submits. Never sent without approval.
Assisted
Software gathers, drafts or surfaces. Judgement and wording stay with the person.
Human only
We do not touch it. Not a roadmap item — outside the product's boundary.
New task
Created by the software: reviewing staged output before it is filed. This is the human gate.
Loading task statements…
This application incorporates information from
O*NET Web Services
by the U.S. Department of Labor, Employment and Training Administration (USDOL/ETA).
O*NET® is a trademark of USDOL/ETA.
Why this page exists
It is a scope boundary, not a pitch. A practice administrator evaluating this
product needs to know what changes about their coordinator's day and what does not. Using the
Department of Labor's task list rather than our own description means the boundary is stated in
a vocabulary neither of us wrote.
It is also the measurement frame. The 30-day ledger measures named tasks from
this list on one payer family — not an abstract percentage. The headline the ledger reports is
cases completed per coordinator per day; the task-level minutes sit beneath it
and explain why the number moved. A desk does not add a seat because a case takes longer — it
adds a seat because the queue will not clear.
And it answers the obvious question about AI. As models improve, more of this
list moves from human to assisted to staged. What does not disappear is the review task — the
person confirming that staged output is correct before it is filed. That task is created by the
software, not removed by it, and it is the one carrying legal accountability for the submission.
Our position depends on that task existing. We would rather state that plainly than pretend the
boundary is permanent.
These task codes are what the measurement reports against. In a 30-day
measurement the ledger records time against the named tasks above rather than against a
private definition of "handling time" — so the result can be checked, and compared with
another desk, without taking our word for anything.
See a case run end to end →
DeskGate
BAA active · Vertex AI · 28 Sep 2026