Coverage appeal AI
Counterforce Health
Counterforce Health is a Durham-based startup offering free AI-generated insurance appeal letters for patients and small clinics, funded by grants from the NIH and the University of Pennsylvania and chaired by CareYaya CEO Neal Shah. Its public commitment that the service will always be free for individuals and will never accept insurance-company money is a strong patient-alignment signal. CAIHL review should keep watching its self-reported 70 to 75 percent success claims, the privacy policy allowance for training AI models on patient data with de-identification only where feasible, and the undisclosed corporate status behind the .org domain.
Public-source research has been drafted; final human publication review and change-log detail are still required.
Summary judgment · 84% toward patient-directed
Potentially agency-expanding, with AI-training and verification caveats
Counterforce gives patients a free, anti-denial advocacy tool and pledges never to take insurer money, but its privacy policy permits training AI models on patient appeal data and its success-rate claims are self-reported.
Patient agency
How this tool changes agency
Appeal letters, appeal-rights guidance, voice AI follow-up, reported regulator notifications, rural Appeals on Wheels outreach, and free 1:1 case management are all direct action support.
Letters are drafts the patient reviews and submits, and the privacy policy grants access, correction, and deletion rights, but in-app editing and challenge workflows were not verified in this pass.
Patient-facing signals
Who does this AI serve?
Built by founders citing their own denial experiences, free for individuals, pledged never to accept insurer money, grant-funded by NIH and the University of Pennsylvania, with future revenue planned from clinic subscriptions rather than patients.
Can patients tell AI is involved?
AI appeal generation and the Maxwell voice AI are the advertised product, and AI involvement is central to the public pitch.
Can patients meaningfully choose?
Use is voluntary, patient-initiated, and free, with free human case management offered as an alternative for complex cases.
Can patients correct or challenge what the AI produces?
Letters are drafts the patient reviews and submits, and the privacy policy grants access, correction, and deletion rights, but in-app editing and challenge workflows were not verified in this pass.
Does it help patients understand or act?
Appeal letters, appeal-rights guidance, voice AI follow-up, reported regulator notifications, rural Appeals on Wheels outreach, and free 1:1 case management are all direct action support.
Text findings
Who is left out or burdened?
Free model lowers barriers, evidence incomplete
Zero cost, rural mobile outreach, and small clinic grants reduce burden, but language support, disability access, low-literacy support, and non-digital channels are not documented.
What happens to patient data?
Disclosed, with AI-training caveat
Collects denial letters, plan details, and medical context, says it does not sell personal data for money, but discloses using data to train and validate AI models with de-identification only where feasible, shares with third-party AI providers under safeguards, displays a HIPAA-compliance badge without covered-entity detail, and retention is open-ended.
Are the clinical boundaries clear?
Partial
The service is administrative advocacy rather than medical advice, but public materials do not spell out escalation paths, uncertainty handling, or safety boundaries for medical decisions tied to appeals.
Who defined what good looks like?
Vendor-defined, with academic ties
Success claims of 70 to 75 percent and a 2x national-average win rate are self-reported, grant-funded research relationships with the University of Pennsylvania and Duke exist, but no published independent evaluation was found in this pass.
Review method
Public-source review of the official homepage, about page, and privacy policy, plus NBC News, Axios, GrepBeat, and CBS17 coverage confirmed via search; no product walkthrough, vendor interview, corporate-records check, or independent verification of success-rate claims.
Draft profile · Medium draft, official sources and press, outcome claims unverified