Detectory

Built for health & public-benefit program integrity

Catch enrollment fraud before it's approved

Is that uploaded paystub genuine? Detectory renders a forensic verdict on every document, then connects the fakes to the synthetic identities and broker rings behind them, with referral-grade audit trails.

Built to the standards public programs require

SOC 2 readinessNIST 800-63 IAL2CMS MARS-ECMMC 2.0StateRAMP
LIVE · Forensic verdict

paystub_march.pdf

APP-48292 · SEP: loss of coverage · broker M. Torres

Template geometrydeviation
Font & kerning integritydeviation
Document metadatareused
87/100
FRAUD · FLAG

Synthetic paystub. SSN active on 3 other policies.

Broker ring linkageconfidence 0.86

14 applications share this template · 3 devices · 1 broker

The reality at the front door

The enrollment front door is wide open

The December 2025 GAO investigation put numbers on what program integrity teams already knew: eligibility verification alone isn't stopping organized enrollment fraud.

$186B

improper payments, FY2025

Up from $162B the year before. The leak is widening, not closing.

Source: GAO

23 of 24

GAO fake applications approved

Undercover filings with fake SSNs and unsupported income sailed through.

Source: GAO, Dec 2025

200K+

unauthorized plan switches

Consumers moved between plans without consent in a single year.

Source: CMS

850+

brokers suspended by CMS

The fraud is organized. It runs through agents and brokers, not one bad form.

Source: CMS

Outstanding Verifications

When trusted data sources fail, the document is all you have

Every state-based marketplace knows the workflow: electronic verification can't confirm income or citizenship, an OV notice goes out, and an applicant uploads a document your team must judge by eye. Detectory answers the one question that queue was never built to answer: is this document real?

A verdict on every upload

Every document submitted to clear an outstanding verification gets a forensic verdict in seconds: genuine, fraudulent, or needs human review. Your OV queue stops being a stack of unanswerable PDFs.

Legitimate enrollees clear faster

Most uploads are genuine. Those verify instantly and coverage continues without a 90-day cliff, without added friction, and without staff hours spent squinting at paystubs.

Federal funds are never paid twice

Cross-program duplicate detection flags the same identity enrolled in both your marketplace and Medicaid, so reconciliation happens before payment, not after an audit finds it.

See every document type we verify, with example verdicts
OV-2026-08841 · income verification uploadForensic scan

EARNINGS STATEMENT

Meridian Staffing LLC · Pay period 05/16 - 05/31

Gross: $2,140.00
Net: $1,782.15
YTD: $23,540.00
  • Font kerning inconsistent with issuer templateAnomaly
  • PDF metadata: created 11 min before uploadAnomaly
  • Employer EIN does not resolve to active entityAnomaly
  • OCR income figures internally consistentPass

Verdict: Likely fraudulent

Confidence 96.2% · routed to human review, not auto-denied

3 of 4 checks failed

Capabilities

From a single fake paystub to the ring behind it

Four detection layers score every application in real time. It starts with a verdict on each document, because a forged paystub is rarely alone: the same template, broker, and identity pattern usually spans hundreds of applications.

Layer 01

Document Forensics & Authenticity Verdicts

A genuine-or-fraudulent verdict on every uploaded document: tamper, template-deviation, metadata, and OCR-consistency signals on IDs, paystubs, tax forms, and special-enrollment letters. The fakes a human reviewer can’t catch at scale.

  • Authenticity verdict
  • Metadata analysis
  • Template deviation
  • OCR consistency
Layer 02

Synthetic-Identity & SSN Intelligence

Fabricated, duplicate, and deceased SSNs. One SSN spread across many policies. We resolve who the applicant really is before coverage is granted.

  • SSA death-match
  • Cross-policy SSN reuse
  • Identity resolution
Layer 03

Broker & Ring Analytics

The fraud is organized. We map the ring behind the applications: shared brokers, reused SSNs, device clusters, and enrollment velocity. Not just one bad form.

  • Broker velocity
  • Device clustering
  • Template reuse graphs
Layer 04

Referral-Grade Audit Trail

Defensible evidence, human-in-the-loop. Every flag survives Congress, CMS, and your MFCU, and you never wrongly cut off a legitimate enrollee.

  • Chain of evidence
  • Reviewer sign-off
  • Export-ready packets

Case workflow

You stay in control. Every action is defensible.

Five stages from monitoring to referral. A human signs off before anyone loses coverage, so your evidence holds up and no legitimate enrollee is ever wrongly cut off.

Level 1

Monitor

Score every application across four fraud layers

Every application entering your marketplace is scored in real time across documents, identity, ring behavior, and eligibility signals. Clean applications flow through untouched.

Live scoring stream

  • APP-48310
    Clear
  • APP-48311
    Clear
  • APP-48312
    Flag
  • APP-48313
    Clear

1,284 scored today · 3 flagged · 0 friction added

Getting started

From application to referral in one pipeline

Three stages, one integration. Here is what your program integrity team actually sees at each step.

01

Ingest

Sits alongside your hub — no rip-and-replace

Applications, uploaded documents, and broker submissions flow in from the marketplace and eligibility systems you already run.

Connected sources

Marketplace / SBEEligibility hubBroker portalDocument uploads
1,284 applications ingested today
02

Score

Four detection layers, in real time, pre-eligibility

Every application is scored before eligibility is determined — across documents, identity, ring behavior, and eligibility signals.

Detection layers

  • Document forensicsauthenticity
  • Synthetic identitySSN / face reuse
  • Broker & ringshared templates
  • Eligibility signalscross-program
Risk 87 · FLAGreal-time
03

Refer

Investigator-ready, human-in-the-loop

High-risk cases route to a queue with a referral packet and full audit trail, formatted for the bodies that hold you accountable.

Referral packet

  • Chain of evidence
  • Reviewer sign-off
  • Export: MFCU · CMS · Council
No enrollee auto-denied

One engine, every regulated front door

Built for health marketplaces. Proven wherever documents decide dollars.

A fake paystub looks the same whether it's attached to a marketplace enrollment, a Medicaid application, a loan file, or an insurance claim. The forensic engine we built for the hardest buyer in the market, a state health exchange, applies anywhere a document decides eligibility or payment.

Built for government

A compliance path your procurement team can trust

Public programs can't buy what they can't certify. Here's exactly where we are and where we're headed. No vague “enterprise-grade security” hand-waving.

Baseline

SOC 2 readiness

Security, availability, and confidentiality controls organized for customer review and audit readiness.

Next

NIST 800-63 IAL2 + CMS MARS-E

Identity-proofing assurance and the ACA exchange security baseline your CISO will ask about first.

Federal contractors

CMMC 2.0

A control roadmap for defense-adjacent fraud programs and contractors handling FCI or CUI.

Scaling

StateRAMP

A state cloud security path for moderate-impact program integrity workloads.

In Progress

FedRAMP

A federal cloud security roadmap for CMS and Treasury-scale programs.

Prove your marketplace is clean

See Detectory score real enrollment fraud, from fake documents and synthetic identities to the broker rings behind them, in a 30-minute walkthrough. No sales pitch, just the product.