For national payers, Medicare Advantage and Medicaid plans, and SIU

The money moved months ago. The proof never did.

The file is oriented to FRE 902(14). A court decides admission. This page does not publish an ROI multiple or a recovered-dollar total. The fee is an ROI-based fee plus a floor license. The plan keeps the recoveries. There is no contingency.

Unsupported payments are almost always found after the fact, then handed to counsel as a spreadsheet. We score the decision before the money moves, and seal the evidence in that same moment, inside your own data platform.

Request an exposure assessment Meet AAVAE, the RADV workbench
Chapter IWhat the number hides

Most of it is not hidden fraud. It is a gate that was never there.

Payments clear first. Documentation gets checked later, if ever. By the time anyone asks whether a claim was supported, the money has moved and the trail has gone cold.

We put the gate back where it belongs: identity, enrollment, exclusions, documentation and policy, verified before release. Not to deny claims. To make sure every decision has its reason attached.

CMS, fiscal year 2024 $87.1B in improper payments across Medicare, Medicaid, CHIP and the ACA marketplaces.
79%of Medicaid improper payments traced to insufficient documentation
$31.7Bimproper in Medicare fee for service alone
Source: HHS Agency Financial Report, FY2024.
Chapter IIWhy good cases fall apart

The fraud was real. The evidence just could not survive discovery.

A spreadsheet is not a case

Counsel inherits a list of flagged claims and months of work: prove where the data came from, that custody held, that nothing was altered. Every step reconstructed under pressure.

A captive witness is discounted

When the audit technology belongs to a competing plan or provider network, its findings carry an asterisk. We own no health plan and run no provider network.

A contingency fee taints the story

A vendor paid from what it finds has a reason to find more. The plan keeps the recoveries. The fee is an ROI-based fee plus a floor license. There is no contingency.

Chapter IIIThe patterns

Every scheme leaves a shape. We seal it while it is still visible.

The provider who was never therePhantom providers billing for care never rendered. Enrollment, identity and claim patterns sealed at capture.
Forty-seven wheelchairs a monthDMEPOS upcoding from a storefront. Coding and utilization vectors packaged with chain of custody.
The hospice capCap and eligibility integrity, sealed as AREB for defense or CREB for recovery.
Pharmacy and 340B arbitrageCross-claim patterns where documentation gaps become evidence bundles.
The RADV sampleCMS-HCC V28 is the payment model for payment year 2026. Earlier payment years keep the model that applied that year. The sealed samples on this site name V22 and V24. V28 is not added into those sealed dollars.
The SIU queueRanked, documented, and shadow by default. The investigator decides; we never adjudicate.
Chapter IVInside your walls

Your data never leaves home.

We come to your data, not the other way around. JIL runs inside your Snowflake or Databricks estate. Charts stay where they are. Only a cryptographic fingerprint ever leaves, enough to prove the record has not changed, and nothing more.

01
Snowflake
Detection and evidence jobs run in your account. No PHI egress for core evaluation.
02
Databricks
Unity Catalog aligned pipelines for claim and encounter validation at scale.
03
Dedicated enclave
An isolated environment for agency and MFCU style deployments.
Chapter VA first proof

Start with one payment class. Let it make the argument.

A narrow pilot of two to four weeks, on one payment class, provider segment or region. Your rails stay exactly as they are. At the end, you hold the findings, and a clear view of what a sealed record would have changed.

Bring us the claim you are worried about. We will show you its story.

A RADV sample, a payment on the desk, an SIU file that is not yet a case. A written view in two business days.

Request an exposure assessment