Coverage appeal AI
Fight Health Insurance
Fight Health Insurance is a free, pay-what-you-want AI appeal generator founded by engineer Holden Karau after her own denial fights, run by Delaware-incorporated Fight Health Insurance, Inc. and sustained by the paid provider-facing Fight Paperwork product. It is unusually transparent for this category, with mostly open-source code, a public page documenting its fine-tuned models and training data, explicit hallucination warnings, and no marketed success-rate percentage. The main CAIHL tensions are a terms-of-service clause permitting model training on submitted information while asking users to strip their own identifiers, and a privacy policy that allows business-partner and ad-tech sharing.
Public-source research has been drafted; final human publication review and change-log detail are still required.
Summary judgment · 90% toward patient-directed
Potentially agency-expanding, with strong transparency posture
Free, mostly open-source, founder-driven appeal generation with on-device OCR and data deletion tooling, though the terms permit model training on submitted data and put the identifier-removal burden on users.
Patient agency
How this tool changes agency
Appeal generation, plain-language denial explanation, policy explanation, a state help directory, and AI chat support both understanding and action across the appeal process.
Strong points include on-device OCR, deletion tooling, a consumer health data notice, open-source code, and a statement that no real patient appeal letters are used in fine-tuning, but the terms permit training models on submitted information, the privacy policy allows business-partner and ad-tech sharing that may legally be a sale, and FHI is not a HIPAA covered entity.
Patient-facing signals
Who does this AI serve?
Free for everyone with optional pay-what-you-want support, revenue from the separate provider-facing Fight Paperwork product, mostly open-source code, and no insurer or pharma funding disclosed.
Can patients tell AI is involved?
AI use, named base models such as MedGemma, training-data sources, energy use, and limitations are documented on a public About Our AI page.
Can patients meaningfully choose?
Use is voluntary and free, with an on-device OCR option that limits data sharing, a choice between self-submission and paid fax delivery, and a dedicated data deletion page.
Can patients correct or challenge what the AI produces?
The tool generates multiple drafts the patient is told to review and edit, warns that models can hallucinate citations, and provides data correction and deletion rights, though workflows were not verified in-app.
Does it help patients understand or act?
Appeal generation, plain-language denial explanation, policy explanation, a state help directory, and AI chat support both understanding and action across the appeal process.
Text findings
Who is left out or burdened?
Low cost barrier, high self-service burden
Free access helps, but users must review and edit AI drafts themselves, the terms put the burden of removing identifying information on the user, and language, disability, and low-literacy support are not documented.
What happens to patient data?
Mixed, transparent but with tensions
Strong points include on-device OCR, deletion tooling, a consumer health data notice, open-source code, and a statement that no real patient appeal letters are used in fine-tuning, but the terms permit training models on submitted information, the privacy policy allows business-partner and ad-tech sharing that may legally be a sale, and FHI is not a HIPAA covered entity.
Are the clinical boundaries clear?
Clear
Repeated statements that outputs are not legal or medical advice, explicit hallucination warnings, and instructions to verify cited literature and consult professionals for complex situations.
Who defined what good looks like?
Founder-defined, open to scrutiny
No marketed success-rate percentage, which is unusually restrained for this category, and open-source code allows outside review, but no independent evaluation of appeal quality or outcomes was found in this pass.
Review method
Public-source review of the official homepage, About Our AI page, privacy policy, terms of service, and the Fight Paperwork about page, plus SF Standard coverage confirmed via search; no product walkthrough, code audit, or vendor interview.
Draft profile · Medium draft, official sources and press coverage