🔍 We publish what we don't have yet. Every claim on this page names its source — and where we're early, we say so. Read why →
⚕ Built for skilled nursing facilities · Starting in New Jersey

The Denial Recovery Platform for Skilled Nursing

From the remittance file to the appeal decision, Vero keeps every denial moving — and never lets a filing deadline die quietly.

Book a 20-minute call See how it works
90-day pilot, free, cancel any time. You pay only on dollars actually recovered.
app.verohealthrcm.com — demonstration with invented data, no live system exists
🔥 Act now 4
📄 Ready to approve 3
❓ I am stuck — teach me 1
⏳ Waiting on payer 14
✓ Recovered
Coverage cut at day 11 — Meridian Advantage$7,140
E. Rosen · Room 214 Appeal drafteddeadline 01/01 — 60 days from the 11/02 determination, 42 CFR 422.582
Medicare records request (ADR)$9,870
R. Sizemore · Room 108 62-page package assembleddue 09/25 — the ADR letter's own date governs
A denial I have never seen before$12,480
W. Brenner · Room 117 I don't know how to handle this — teach me
Medicaid application — state wants two documents$29,513
L. Hobbes · Room 302 Family email drafted — awaiting your authorization check
Every name, dollar and result above is invented for demonstration. We have no customers and no live system yet — see the honest part.
95%of appealed SNF denials were overturned
18%of denials were ever appealed at all
12%of SNF admission requests were denied by MA plans
Source: HHS Office of Inspector General, report OEI-09-24-00331 (June 2026), sampling June 2024 data across 19 Medicare Advantage organizations. These are sample findings about an industry — not a probability for any single case, and not our results. We have no results of our own yet.
The problem

Denials still run on faxes, sticky notes, and memory

A Medicare Advantage plan cuts a resident's stay at day 11. Your business office is one person covering five payers. A proper appeal — record pull, regulation cites, point-by-point rebuttal — is most of an afternoon that person does not have. So the claim gets written off, and the federal data above says most of those write-offs would have been won.

The platform

One queue. Every denial. Ranked by money and time.

Vero reads the paper you already receive, computes every clock in code, and writes the first draft — your staff reviews and clicks send. Nothing leaves without a person.

Never lose an appeal to a dead deadline

Every filing clock — the 60-day reconsideration, the ADR response window, the noon-before-the-effective-date QIO call — is computed by ordinary code from a rules table where every entry carries its regulation and the date a human last checked it. The AI writes prose; it is never allowed to compute a date or a dollar.

  • Deadlines shown with their arithmetic and citation, so you can check us
  • Escalation ladder as a deadline approaches — it cannot be quietly missed
  • The deadline calendar lives outside the AI: if AI service fails, your worklist and clocks survive
Appeal deadline — E. Rosen⏰ Jan 1
CALCULATED: 60 days from the 11/02 determination (42 CFR 422.582). Determination date READ from the plan's letter, page 1.
QIO fast-track window⏰ noon, Nov 4
Noon of the day before the 11/05 effective date (42 CFR 405.1202(b)) — not "the day after receipt"; that is the hospital rule.
ADR response — R. Sizemore⏰ Sep 25
The window printed on the ADR letter itself governs. Miss it and a payable claim becomes an automatic denial (code 56900).

Appeals written from your own chart, not a template

The draft quotes the plan's exact denial sentence and rebuts it point by point — with the physician order, the MAR count, the wound log, each cited to where it lives in your record. Facts the software cannot verify are asked, not asserted: it will tell you a signature is present and dated; it will never tell you a signature is valid.

  • Cites 42 CFR 409.31, CMS-4201-F and Jimmo where the record supports them
  • Every fact carries its source — a fact without provenance does not render
  • Judgment calls (POA scope, note sufficiency) block until a named human confirms
"The member no longer meets criteria for skilled level of care… is noted to be self-feeding and requires minimal assistance with hygiene and grooming."
Draft rebuttal, point 2for your review
"Self-feeding and grooming independence are not coverage criteria. No provision of the Medicare statute, regulations, or manuals conditions SNF coverage on ADL dependence."
READ: IV ceftriaxone q24h — physician order 11/01 + 12 administrations in the MAR · wound log 11/01–11/12 · MD note 11/03, quoted verbatim.

Your morning becomes a ten-minute review

Overnight it reads the remittance lines and letters that arrived, re-sorts the queue by dollars × days-to-deadline, and leaves drafts waiting. The morning brief says what needs a decision today — everything else is already moving.

  • Reads 835 remittance copies, forwarded letters, faxes and portal PDFs
  • Nothing is re-routed — your existing auto-posting keeps working untouched
  • Weekly report shows raw counts, never a percentage without its denominator
What we did while you were asleep
Read 47 remittance lines and 3 letters · recalculated every deadline · wrote 3 documents for your approval · hit 1 denial we could not classify — and stopped instead of guessing.
Our overturn rate—
"We do not publish a number until we have enough of your own decided appeals to be honest about it. 11 of your 18 are still open."
When it doesn't know

The most important screen says "I don't know"

Ask any AI vendor what their system does with something it has never seen. The honest answer matters more than the demo — it is the moment the software either protects your money or quietly loses it.

It stops, shows its work both ways, and asks to be taught

An unfamiliar denial gets a plain first sentence — "I do not know how to handle this one" — then two lists: what it could read (each item with its source) and what it could not determine and why. Then four questions for your biller. One answer is enough; what she says becomes a written note with her name and date on it, marked 1 case, unproven until it actually works.

  • No model retraining — it writes notes you can read, edit, or delete; delete one and the behavior reverts
  • Never learns a deadline or dollar rule from habit — those come from regulation and your contract
  • Nothing learned in your building ships to another facility
  • The deadline keeps running and escalating while it waits to be taught
? I do not know how to handle this one.
CO-B15 / N674 — "plan Transition of Care assessment not on file." I searched the chart index and fax log and did not find a document with that name. That means I did not find one — not that one was never done.
Closest match in my library: CO-197, no-prior-auth. It is not the same denial — reusing that playbook would look competent and be wrong, so I have not.
Teach me: what is this, in your words? What's the first thing you do?
Security & compliance

Stated plainly, because you'll ask

No vendor can honestly promise "unhackable." We promise layered healthcare-grade controls, a strictly limited blast radius, and a written status for every control — in place, before your data moves, or planned. We never blur the three.

🔏 HIPAA BAA signed before any record moves
🇺🇸 AWS HIPAA-eligible, US regions stood up before pilot
👤 Human approval on all outbound day-one design rule
🛡 SOC 2 & pen test roadmap — not started

Full control-status table, subprocessor list and BAA draft available on request. We'd rather lose a deal than misstate this.

The honest part

You'll notice there are no customer logos here

Vero is Latin for true. We named the company after the one rule we will not break: nothing appears in front of you as a fact unless we can name where it came from — the document, the regulation, the date. That rule is why there are no customer logos on this page.

We don't have customers yet. No testimonials, no case studies, no "trusted by" carousel — we are an early-stage company looking for our first design partner in New Jersey, and we think you should hear that from us rather than discover it.

What we do have: a knowledge library built from ~80 primary sources — decided appeal cases, CMS manuals, MAC guidance, the OIG's audit work — and a working discipline we call the Truth Standard: nothing appears in front of a customer as a fact unless we can name where it came from. Every deadline on this page carries its regulation. Every statistic carries its report number. Where we were wrong, we keep a public register of what we fixed.

What a design partner gets: the pilot free, our full attention, pricing locked for two years, and a product shaped around your building's actual denials. What we get: the thing no amount of research can substitute — reality.

Pricing

We get paid out of money you weren't collecting

If we recover nothing, you pay nothing. That's the entire pricing conversation.

15%
of denied dollars we help you recover
  • No setup fee, no platform fee, no minimum
  • 90-day pilot at zero cost, cancel any time
  • You pay only on dollars that actually land
  • Pilot pricing locked for two years if you continue
  • Monthly report of exactly what was recovered and how
Questions we always get

The honest answers

Do we have to change our billing system?

No. Vero reads copies of what you already receive. We do not re-route your remittance advice, so auto-posting in PointClickCare or MatrixCare continues untouched. Setup is adding a delivery copy at your clearinghouse and forwarding an email address — and we will confirm the exact enrollment steps with your clearinghouse before we promise you a timeline, because we have not run that process with them yet.

Does the AI send things without us seeing them?

Not in normal operation. Every outbound appeal, email and records request waits in an approval queue until a person on your team reviews it and clicks send. As you build confidence you can choose to automate low-risk routine items — your decision, reversible.

What happens when a denial arrives that it has never seen?

It says so — in the first sentence — and does not write a letter. It shows what it could read with a source for each item, what it could not determine and why, and asks your biller four questions. One answer becomes a written note with her name on it, and the next one of those gets drafted instead of stopped. The deadline keeps running and escalating the whole time. Expect this in your first weeks; that is the system working, not failing.

Does it learn from us? Are you training AI on our data?

It learns, and no. We do not fine-tune models on your claims or residents. It writes notes — visible on one screen with an author, a date and a case count. You can read, edit, or delete any of them; deleting one reverts the behavior immediately. Nothing learned in your building goes to another facility, and it never learns a deadline or dollar rule from habit.

What if the AI writes something wrong?

The approval step means the cost of a bad draft is a bad draft, not a bad filing. Beyond that: every factual claim in a letter must resolve to something in your record before the draft is shown, and dates and dollar amounts are calculated by ordinary code, never by the AI.

How do you know an appeal is worth filing?

On three things we can see: the dollars, the days left on the filing clock, and whether your records contain the evidence the regulation requires. We will not pretend to know payer win-odds before we have decided cases — and sometimes the honest answer is "don't fight this one," and we say so.

You're new. Why should we take the risk?

Fair. The pilot is free and cancellable, we sign a BAA and stand up the security controls before touching data, and we never gain the ability to move money or submit claims. The realistic downside is an hour a week for ninety days. The upside is the 82% of denials in the OIG's sample that nobody ever fought.

Book a 20-minute call

Bring three months of remittance history and we'll walk through what was denied, what was never appealed, and what we believe is still recoverable — in writing, no charge, no commitment.

Book the call →