Billers stopped typing their own numbers
The billing dashboard should run on the claims themselves, not on what someone remembered to type.
- Service
- Revenue cycle
- Status
- in production
- Built
- October 2026
2,727
claims captured in the first history load, across 72 biller-days
Source: Build notes, Oct 2026
58 / 52
rejections found already accepted / still open on the first follow-up pass
Source: Build notes, Oct 2026
The problem
The billing dashboard depended on numbers each biller typed in at the end of the day. Typed numbers go stale, drift from the truth, and are easy to dispute. Rejections were followed by hand.
What was built
Every 30 minutes during business hours, an RPA bot reads the claims each biller sent from the EHR, in every status, plus the clearinghouse's claim-status reports. It links each rejection back to its claim (or, failing that, the patient) and to the biller. The billing dashboard runs entirely on that data. Each biller gets a daily performance email and a Monday email with their open rejections, with managers copied; managers get a weekly summary.
Providers with notes still open or missing seven or more days after a visit get a weekday 8 AM email listing them, using the marks billers already make on the worklist, so no separate list is needed.
What was hard
A short or partial read could have wiped out a day's claims, so a read that returns less than 60% of a day's known claims no longer removes anything. Rejection status is followed only from the EHR, so each rejection moves to accepted, re-sent or rejected again on its own.
Result
The first history load captured 2,727 claims over 72 biller-days, at about seven seconds per biller-day. The first follow-up pass found 58 rejections already accepted and 52 still open.
Where the numbers come from
Build notes from October 2026.