A 400× gap between how often patients win and how often they try. The work — deadlines, claim-file requests, medical-policy citations, appeal letters — lands on patients during a health crisis, so almost no one does it.
nH Predict, PXDX, and EviCore turned denials into predictable, attackable templates we can classify and rebut at scale.
The 2024 Senate investigation and 2025 CMS rule changes standardized notices and put insurers on the defensive.
What used to take an advocate hours — drafting a cited appeal — now costs cents per case, so the unit economics finally work.
Photo, PDF, or email-forward the denial letter.
Identify denial type and surface procedural defects.
Draft a citation-backed appeal in the right template.
Submit through the right channel — fax, portal, or QIO.
Watch every deadline; escalate if denied again.
Powered by a proprietary, outcome-labeled denial corpus and a classifier + template library that improve with every case.
Modeled on conservative funnel and pricing assumptions; contingency and subscription compared in the financial model.
Year-1 burn ≈ $382K, then profitable as volume ramps. All figures are model assumptions — see the financial model.
Ship the MA post-acute flow end to end.
California, New York, Michigan — strong external-review programs.
Target ≥1.5× overturn rate vs. unaided appeals on the first 100 cases.
Each case enriches an outcome-labeled corpus + template library.
| DIY / generic AI | Patient advocates / attorneys | Overturn | |
|---|---|---|---|
| Speed in a 72-hr crisis | Variable | Slow / scheduling | Minutes |
| Cost | Low | $$$ hourly / contingency | Flat per-appeal |
| Denial-specific expertise | Generic | High | High + automated |
| Deadline & escalation tracking | None | Manual | Built-in |
| Scales to millions of cases | No | No | Yes |