
A North Carolina man denied a medically necessary drug got it approved five days after letting artificial intelligence write his appeal letter.
Story Overview
- A patient’s insurance appeal, drafted with help from the AI tool Claimable, led to approval of his Remicade prescription within days.
- Startups like Claimable and Counterforce Health build custom appeal letters using medical records, policy language, and past successful cases.
- Fewer than 1% of denied health insurance claims get appealed, even though insurers often reverse their own decisions when challenged.
- Questions remain about accuracy, privacy, and whether insurers will respond by building tougher AI defenses of their own.
The Approval That Caught the Country’s Attention
TIME magazine reported in November 2026 that a patient used Claimable to fight a denial for Remicade, an infusion drug for inflammatory conditions. The AI tool built an appeal around his missed workdays and cited federal statutes requiring coverage of medically necessary treatment. Five days after submission, his doctor’s office confirmed the drug was approved. The story captured a growing shift: patients using software instead of lawyers to push back on insurance companies.
How The Software Actually Builds a Case
These are not experimental lab projects. NBC News reported that Claimable produces personalized letters pulling from clinical research and the outcomes of other patients’ appeals for similar denials. Counterforce Health, one of the companies building these tools, says its system reviews a person’s policy terms and matches them against appeals that already succeeded for the same treatment. The goal is simple: turn a complicated insurance fight into a document insurers cannot easily wave away.
Cost has also dropped sharply. PBS NewsHour reported that these tools ask patients for the documentation insurers typically expect, then generate a full appeal for around $40 to $50. Some versions are free. For a process that once required hours with a lawyer or benefits advocate, that price tag changes who can realistically fight back when a claim gets rejected.
Why So Few People Ever Push Back
The real problem these tools target is participation, not persuasion. Research from KFF found that in 2024, insurers on the Affordable Care Act marketplace denied about 19% of in-network claims, yet consumers appealed fewer than 1% of them. Of those who did appeal, insurers upheld their own denial a majority of the time, but a meaningful share got reversed, proving the system often works for people willing to fight it.
That gap has existed for years. A 2017 KFF analysis found an average appeal reversal rate of 14%, with individual insurers ranging anywhere from 1% to 88% depending on the company. Most people simply give up, overwhelmed by paperwork, medical jargon, and strict deadlines. AI tools exist to close that gap by doing the tedious work: organizing records, matching language to policy rules, and producing something an insurer has to seriously consider.
What Still Needs Answers
No large study has yet proven exactly how often AI-assisted appeals succeed compared to traditional ones, and some reporting has raised concerns about insurers using their own AI systems to screen out unlikely appellants in the first place. Patients handing over medical records and personal health details to a new software company also deserve clear answers on how that data gets stored and used. Those are reasonable questions, not reasons to dismiss what is already working for real patients today.
For now, the practical lesson is straightforward. Insurance denials are common, appeals are rare, and the rare appeals that do get filed often succeed. Tools that make filing faster and cheaper remove the biggest barrier stopping patients from trying. Whether someone uses a $40 app or writes the letter themselves, the data says one thing clearly: silence after a denial almost always favors the insurance company, not the patient.
Sources:
time.com, wheninyourstate.com, counterforcehealth.org, acsh.org, pbs.org, kff.org













