Skip to content

Agentic AI for False Claims Act qui tam litigation

Backed by

Y Combinator

Introducing the Erinys Engine

The case engine for False Claims Act work

Seven stages that build a qui tam case stone by stone — from the first relator interview to a filing the government wants to join.

  1. 01

    Security & Privilege

    The foundation everything rests on. Sealed by statute, secured by design — work-product protection, encryption, and zero training on your data.

  2. 02

    Intake & Triage

    Every tip screened against public data, prior filings, and the first-to-file bar. The cases worth building get built — fast.

  3. 03

    Investigation Agent

    An agent that plans the investigation, probes claims data and contracts, and corroborates the relator before a single hour is billed.

  4. 04

    Evidence & Integrations

    CMS, USAspending, SAM.gov, PACER, and the relator’s own documents, normalized into one evidence graph with provenance on every fact.

  5. 05

    Damages Modeling

    Overpayment estimates, sampling, and treble exposure modeled claim by claim — materiality answered before the government asks.

  6. 06

    Drafting

    Complaint and disclosure statement drafted from the evidence graph, every allegation pinned to its exhibit, Rule 9(b) particularity built in.

  7. 07

    DOJ Submission

    The capstone. Disclosure packages assembled around what line AUSAs actually read, so the government is handed a case it wants to join.

Every fact, filing, and decision in a qui tam case — connected in one engine.

↳ See it on a live case — book a demo

Platform highlights

Built for the whole life of a case.

From first tip to filed complaint, every product works the same case graph.

Investigator

Runs the case end to end — planning the investigation, pulling evidence, drafting the filings. Every step is reviewable, and the judgment calls stay yours.

Signals

Standing watch over CMS, SEC, USAspending, and procurement data. Anomalies are flagged, scored, and checked against first-to-file exposure before anyone else notices them.

Ledger

One record of every claim, witness, and exhibit — provenance attached, so the case graph and the disclosure statement never drift apart.

War Room

Bring relators and co-counsel into the case without surrendering control: privilege-aware access, granted artifact by artifact.

Solutions

Every scheme. Every sector.

Take billing fraud to task

Medicare Advantage risk adjustment is DOJ’s stated priority — unsupported diagnosis codes drove a record enforcement year. Erinys cross-references HCC coding against encounter documentation at scale, and still catches the classics: upcoding, unbundling, kickbacks, and medically unnecessary care.

The market

The FCA is the government’s primary fraud weapon.

Qui tam filings hit records while relator’s counsel still work with tools built for a slower era. 77% of FY2025 recoveries began with a relator. The firms that industrialise investigation will own the docket.

$6.8B
recovered under the False Claims Act in FY2025 — the largest year in its history
1,297
qui tam suits filed in FY2025 — an all-time record
$2.3B
recovered in declined cases, where relators litigated alone

Source: U.S. Department of Justice, Civil Division — FY2025 fraud statistics (Jan 2026).

Security

Built for the confidentiality of sealed cases.

Qui tam work happens under seal. The platform treats that as a first-class engineering constraint.

Privilege-aware by design

Work-product boundaries are enforced in the data model, not by policy documents.

No training on your data

Case files never train models. Yours stays yours — contractually and architecturally.

Encryption everywhere

In transit and at rest, with per-matter isolation and customer-managed key options.

Audit-ready logging

Every access and every agent action is recorded. Produce the trail when it matters.

Discover the future of qui tam work through a live demo.

See the Erinys Engine in action — from intake triage and agentic investigation to a filing-ready disclosure statement.