Denial management, from zero.

After a clearinghouse change, payments landed in AdvancedMD with no matching patient or visit. The billing platform had no denial filter at all, so the only way to find a denial was to open every explanation-of-benefits report by hand, while appeal windows were closing.

I asked the billing specialist which denial codes to look for. There were too many to list. But the specialist could name, from memory, the five codes that mean a claim was approved.

So the filter works the other way around: it lets those five through and surfaces everything else for review. A code nobody has seen before lands in the queue instead of disappearing.

A parser that reads every report and reconciles it against our claims to the penny.

• 539 reports and 1,291 claims parsed, with nothing lost
• A shared work queue where the billing team marks each claim worked, and every mark survives each refresh
• A default-deny connection to the billing platform: submitting claims and posting payments are refused in every mode

$169,190 in payments sitting unposted across 434 checks, and about $50K that payers had underpaid. 29 claims had been marked denied by the payer in a field nobody was reading.

With denials finally visible, the root cause showed up. Across 1,740 denied claims, 47.7% traced back to intake, not billing. The fix was one eligibility check at order entry.

The rescued clinical archive now backs 86% of active denial appeals with the report that proves the service was done.

With my agentic team, from the first conversation with the billing specialist to a working queue, tested against the numbers that matter: every report parsed, every claim accounted for. HIPAA is part of the design: patient data minimized, access role-based at the API layer, and a person reviews before anything goes live.