Press & Newsroom

An AI for insurance appeals that won't cite a source that doesn't exist.

Merits turns a denied claim into a complete, source-cited appeal in about a minute. It works only from a closed library of real payer policy, federal law, and review decisions, and quotes each one with a link — so nothing in the letter is made up. It's built for the solo doctors and small practices that would otherwise write the denial off.

Press contact[email protected]
News

For immediate release · June 2026

A tool that writes insurance-denial appeals for doctors — and won't cite a source that doesn't exist.

Merits turns a denied claim into a complete, signature-ready appeal in about a minute, with every argument quoted from the payer's own policy and federal law. $9 a letter. No account, no contract.

June 2026 —Most denied medical claims are never appealed. Usually not because the denial was right, but because appealing one by hand takes an hour a small practice doesn't have. The money gets written off, and the patient often eats the bill.

Merits is built to remove that hour. A physician or biller uploads the denied explanation of benefits; about a minute later they get a complete appeal — the cited letter, the records to attach, and where and when to file it — ready to review, sign, and send.

What sets it apart is what it refuses to do. General-purpose AI will write a confident appeal that cites a regulation that doesn't exist. Merits works from a closed library of the actual rules a denial is won on — 484 payer medical policies across 43 insurers, federal appeal law, Medicare coverage rules, 240 FDA labels, and more than 40,000 external-review rulings — and quotes each source word for word, with a link to open it. If a claim isn't backed by a real source, it doesn't go in the letter.

The patient's name, ID, and date of birth are stripped in the browser before anything is sent. There is no account required to start, nothing to install, and no contract. A single letter is $9, dropping to about $5 in volume.

The gap it targets is wide. When providers appeal Medicare Advantage prior-authorization denials, KFF found 80.7% are overturned — yet only 11.5% are ever appealed. Most denied care is never contested, and most contested care gets paid.

“A winnable denial gets written off for two reasons: the time it takes, and the risk of signing your name to a letter with a citation you didn't check,” said João Guilherme Franco, who founded Merits after two decades building data and product businesses — including inside insurance and anti-fraud — and a decade researching healthcare innovation. “We wanted something that does the hour of work in a minute and can't bluff — every line traces back to a source you can open.”

Multimedia & contact

See it work: a finished, source-cited appeal · fact sheet & brand assets · verification, step by step.
Media: [email protected].

###

What's different

01

It can't invent a citation.

A general AI will write an appeal that cites a regulation that doesn't exist, on a letter a physician signs. Merits draws only from a closed library of real sources and quotes each with a link to the original — if there's no source behind a claim, the claim doesn't appear.

02

The bottleneck was never winning. It's filing.

When providers appeal Medicare Advantage prior-auth denials, KFF found more than 8 in 10 are overturned. The trouble is that almost no one files — each appeal takes an hour a small practice doesn't have, so the denial is written off. Merits removes the hour, not the merits.

03

$9 a letter, not a contract.

Large systems run appeals departments. A two-person practice runs on whatever time is left after clinic. Merits is the per-letter version of that department — no account, no integration, no contract — for the offices that currently eat the denial.

Story ideas
01

The bill small practices give up on

The denials bottleneck isn't winning, it's filing: winnable appeals become write-offs because the hour to draft one costs more than the denial. Merits makes the hour a minute, for $9.

Practice management · healthcare business

02

The health AI built so it can't hallucinate

Most health AI bolts safeguards on afterward. Merits made fabrication structurally impossible — a closed library, every quote re-checked against its source. Verifiable by construction, not by promise.

Tech / AI · health IT

03

The biller's highest-ROI, most-skipped task

Billers know appeals win, but can't justify the hour per claim. At $9 a letter the revenue-cycle math changes, and the appeal stops being the task that's always deferred.

Revenue cycle · medical billing (RCM)

04

What 40,000 overturned denials reveal

Drawn from tens of thousands of external-review decisions: what actually gets care paid on appeal. Ecosystem analysis, not an outcome claim — recurring, and renewable. (Report in progress.)

Data journalism · health policy

05

The health AI designed never to see the patient

In a data-hungry sector, Merits strips patient identifiers in the browser and sells nothing — revenue is the product itself (a letter, or an optional plan), never your data. Privacy by design, incentives aligned.

Health-data privacy / security

06

Prior authorization, from the patient's side

As prior-auth tightens, the filing burden stays the bottleneck. Merits puts the appeal within reach of the practice that currently gives up. (Pitch angle — verify any regulatory specifics at the time of pitch.)

Health policy

See it work

This is the real product, on a sample case (no patient data). Tap any underlined line in the letter — the exact source it rests on opens beside it.

Lakeside Neurology AssociatesFirst-level appeal

RE: claim ANT2026041387721 · CPT 70553 · denial CO-50

Dear Appeals Reviewer:

Lakeside Neurology Associates formally appeals the denial of claim ANT2026041387721 and requests reconsideration. The MRI of the brain performed without and with contrast (CPT 70553) was denied on April 10, 2026 as not medically necessary (CO-50, RARC N115). We respectfully request that Anthem overturn the determination, reprocess the claim, and remit payment for CPT 70553 at the applicable allowed amount. This appeal is submitted within the plan's applicable ACA appeal window, and the exact filing deadline should be confirmed against the denial notice.

Tap any underlined claim — its verbatim source appears →

Coverage policy (LCD)
Verbatim from source
reserved for the patient whose presentation indicates a focal problem or who has experienced a significant change in symptomatology

Why this was cited — Medicare coverage criteria for CPT 70553, cited as persuasive evidence of the accepted standard of care.

LCD L37373 — MRI and CT Scans of the Head and Neck (Noridian)Open the source ↗

[PATIENT NAME] is filled in on your device — it never reaches our servers.

Open the interactive demo →

Product, in use — desktop & mobile

Merits runs in any browser, with nothing to install — and the same prepared appeal works on a desktop or a phone.

Merits on desktop: a prepared appeal showing the amount to recover and the denied claim, ready to review and sign.
Desktop — the prepared appeal, ready to review and sign.
Merits on mobile: the same prepared appeal on a phone.
Mobile — same product, on a phone.

Download full-length captures: desktop · mobile · no patient data (sample case).

How we keep the citation count at zero

Merits never writes from memory. For every claim, it retrieves a passage from the closed library, quotes it verbatim, and then re-checks that quote against its source before the line is allowed into the letter. If the source doesn't say what the claim needs, the claim is dropped — not softened. The closed library is the reason it can't invent a citation; the re-check is the reason it can't misquote one.

The data

A closed library of the rules a denial is won on.

484
payer medical policies, across 43 insurers — quoted back to the same payer
40,000+
external-review rulings, studied for what overturns a denial
1.7M
coding rules — ICD-10, NCCI, and HCPCS
240
FDA drug labels, plus federal appeal law and Medicare coverage rules
0
invented citations — the closed library structurally can't produce one

What it does with that library

~1 min

to turn a denied claim into a ready-to-file appeal — about an hour of work by hand

$9

per letter — or a $20/mo plan; no contract either way

50 + DC

external-review jurisdictions mapped, each with its own rules and deadlines

31

denial-code playbooks, from CO-50 medical necessity to timely filing

The gap it's aimed at

7.7%

of Medicare Advantage prior-authorization requests were denied in 2024

11.5%

of those denials were ever appealed

80.7%

of the appeals that were filed got overturned

Source: KFF, Medicare Advantage prior-authorization data, 2024 (published January 2026). These figures describe the appeal landscape, not a Merits result.

Data reports

We're preparing a recurring analysis — what the external-review record shows about why care gets paid on appeal — built from the review decisions in the library, with published methodology. It analyzes the ecosystem, not Merits outcomes. Reporters who want an early look: [email protected].

Media resources

Founder bio — copy-paste

Short

João Guilherme Franco is the founder of Merits. A product and data executive, he has spent two decades building businesses inside insurance and anti-fraud — industries where the product is trust itself — alongside a decade of research into healthcare innovation.

Medium

João Guilherme Franco is the founder of Merits. He spent two decades building and taking products to market, much of it in the world Merits operates in: he was Marketing, Product & Inside Sales Director for Michelin's connected-fleet (IoT) business across Latin America; built the marketing and data-science function from the ground up at a BTG/Caixa insurer; and led marketing at an anti-fraud leader now part of Serasa Experian, where the product itself was trust and verifiable data. Running under all of it is a decade spent on healthcare innovation.

Long

João Guilherme Franco, founder — a product and data executive who has spent two decades building and taking products to market, much of it in the exact world Merits operates in. He was Marketing, Product & Inside Sales Director for Michelin's connected-fleet (IoT) business across Latin America; before that he built the marketing and data-science function from the ground up at a BTG/Caixa insurer, and led marketing at Brazil's anti-fraud leader, now part of Serasa Experian, where the product itself was trust and verifiable data. Running under all of it is a decade spent on healthcare innovation — graduate research that analyzed more than 100 health technologies, and Health Shift, an ongoing body of work on why modern medicine breaks down and how to rebuild it. Merits is where those threads converge: someone who has built revenue engines and trust products inside the insurance and anti-fraud industries, aimed at the winnable denials small practices and doctors quietly write off.

Brand & downloads

Navy #0C1626Brass #C79A52Teal #84A8A8Paper #FFFDF8Canvas #F5F1E8Ink #16171C

Type: a serif display face for headlines, a sans for body, mono for labels. Keep clear-space of at least the height of the mark on all sides; never recolor, stretch, or place the mark on a busy background.

One-page fact sheet (print / save as PDF)

Need a founder headshot, a vector (SVG) logo, or a different format? [email protected].

A denials primer

Background on the subject itself — sourced, and useful whether or not the story is about Merits.

Denial codesWhat each CARC/RARC denial code means, and whether it's worth appealing.The appeal processInternal appeal, peer-to-peer, and escalation, by plan type.External review by stateWhere a denial goes when the plan still says no — all 50 states and DC.Appeals by payerHow the major insurers' rules, deadlines, and portals differ.Appeals by specialtyHow denials and appeals differ from one specialty to the next.Coverage by drug & conditionGLP-1s, mental-health parity, and other high-volume coverage fights.GlossaryPlain-language definitions for the terms in a denial and an appeal.
FAQ
How is this different from asking ChatGPT to write an appeal?
A general AI produces a fluent appeal that may cite a policy or regulation that doesn't exist — and a physician signs it. Merits retrieves from a closed library of real, current sources, quotes each one, and re-checks every citation against its source before it appears. If there's no real source, the claim is dropped. You can open any citation and read the original.
Does it work? What's the success rate?
The real variable isn't whether appeals win — it's whether they get filed. When providers appeal Medicare Advantage prior-auth denials, KFF found 80.7% are overturned (2024); only 11.5% are ever filed. Merits doesn't claim to move the overturn rate. It attacks the 88.5% that never get written, by turning an hour of work into about a minute. We don't publish a Merits-specific success rate, because it would say more about which denials people choose to appeal than about the tool.
Is it just an AI guessing, or is there something underneath?
It isn't a chatbot writing from memory. Merits runs over a closed library — 484 payer policies, federal appeal law, Medicare coverage rules, 240 FDA labels, and more than 40,000 external-review rulings — retrieves the sources that fit the specific denial, quotes them word for word, and verifies each citation against its source. The retrieval and that citation discipline are the product. The clinical judgment and the signature stay with the physician.
What about patient privacy?
The patient's name, ID, and date of birth are stripped in the browser before anything is sent, so Merits never receives who the patient is. The clinical facts of the case are what build the argument; the identity stays on the practice's device.
Does it need an account, an integration, or a contract?
None of the three. No account to start, nothing to install, no contract. It's a web page: paste the denial, get the appeal. Pricing is per letter.
What can't it do?
It isn't legal or medical advice and it doesn't guarantee an outcome. A human reviews, edits, and signs every letter under their own name. And it won't manufacture a case that isn't there — a plan's flat contractual exclusion, for example, isn't a medical-necessity appeal, and Merits says so.
How does Merits make money? Is there a data business?
Revenue is the product itself: $9 a letter, or an optional $20/mo plan — charged for the appeal, never for data. No ads, no data sold or licensed, and because patient identifiers are stripped in the browser, there's nothing on the back end to monetize even if we wanted to.
Available for comment

João Guilherme Franco is available to comment on: why AI hallucination is uniquely dangerous in documents a clinician signs; the economics of denied claims and why winnable denials get written off; what external-review data shows about why care gets paid on appeal; the prior-authorization burden on small and independent practices; and building verifiability into health AI.

— João Guilherme Franco, Founder · LinkedIn

Media contact

[email protected]

Reaches the team directly.

Boilerplate

Merits is a US tool that drafts insurance-denial appeals for physicians, independent practices, and the billing companies that work their claims — each one quoted from a named, openable source. Per letter; no account or contract.

Founded 2026. meritsappeals.com.