Live Webinar · Aug 10 Gen AI for In-House Counsel: how we post-train an LLM to run Document Review at under 5¢/document. Save Your Seat →
Case Study · White Collar Crime

1 million documents. One search.

Gregor Wynne Arney (GWA) PLLC faced a $35M healthcare fraud and anti-kickback investigation with over one million documents to process. With DecoverAI's embedded ingestion pipeline and custom corpus architecture, the entire document set was searchable and queryable in days—not weeks.

Book a demo →
Matter
Healthcare Fraud
Documents
1M+
Case Value
$35M
Engagement
Custom ingestion build
By the numbers

Scale that would break a standard platform

1M+
Documents processed
Over 1 million documents ingested, classified, and made searchable in a fraction of the time a managed review would require.
$35M
Case value
A $35M healthcare fraud and anti-kickback matter—one of the most complex document review challenges a firm can face.
7
Days to full corpus
From raw document dump to a fully searchable, queryable corpus—in days, not the weeks traditional platforms would require.
01The Brief

Case summary

Gregor Wynne Arney (GWA) PLLC is a white collar crime practice whose clients face existential stakes. When GWA was retained on a $35M healthcare fraud and anti-kickback investigation, the document set that came with it was staggering: over one million documents spanning billing records, clinical notes, communications, contracts, and regulatory filings.

Standard e-discovery platforms aren't built for this. The ingestion time alone would consume weeks of the matter timeline, and the resulting corpus would be difficult to query with the precision a healthcare fraud case demands. GWA needed something purpose-built.

DecoverAI's embedded team built a custom ingestion pipeline designed around the specific document types, metadata structures, and query patterns the investigation required. The result: a fully searchable 1M+ document corpus that GWA's attorneys could interrogate directly—with a single question—from the earliest stages of the matter.

02Why GWA Chose DecoverAI

Four requirements no standard platform could meet

GWA evaluated their options against four non-negotiable criteria. Standard platforms failed on all four.

  • Scalability
    The volume was beyond what any standard platform handles well. 1M+ documents required custom ingestion architecture, not a generic upload-and-search flow.
  • Enhanced accuracy & searchability
    Healthcare fraud involves highly technical documentation: billing codes, medical records, contracts, communications. The platform needed to understand this material and surface it precisely.
  • Data security
    A federal healthcare fraud investigation demands uncompromising security. Patient records, financial data, and privileged communications require enterprise-grade isolation and protection throughout.
  • Precision beyond industry standards
    GWA needed to find specific facts across millions of documents without false positives burying the relevant material. Standard keyword search wasn't sufficient at this scale.
03The Custom Pipeline

Built for the matter. Not off the shelf.

DecoverAI embedded with GWA to build a custom ingestion pipeline designed around the specific document types, metadata structures, and query patterns the case required. The embedded team didn't hand GWA a manual and a login—they built the system alongside GWA's attorneys and then remained available throughout the matter to refine it as the case evolved.

The result was a corpus that GWA's attorneys could interrogate directly. A question that previously required hours of review across thousands of documents on multiple e-discovery platforms could now be answered with a single precise query.

That's not a marginal efficiency improvement. It changes how the firm works: attorneys spend their time on legal strategy, not document retrieval. And they arrive at key facts earlier in the matter—when it matters most.

04What DecoverAI Built

Architecture purpose-built for healthcare fraud

Custom ingestion architecture

The pipeline was built around the specific document types in this matter: billing records, clinical notes, communications, contracts, and regulatory filings—each classified and indexed for the queries GWA needed to run.

Healthcare-specific understanding

The system was trained to parse medical billing codes, procedure classifications, and clinical language—making searches for fraudulent billing patterns precise rather than approximate.

Precision querying at scale

A single natural-language question surfaces the relevant documents across the full 1M+ corpus. No keyword iteration, no platform-hopping, no manual page-turning.

Embedded team support

GWA didn't operate alone. The DecoverAI team remained embedded throughout the matter, tuning the system as the case developed and new document sets arrived.

05The Impact

Command of a million-document record

  • 1M+ documents processed and searchable in days, not weeks—giving GWA command of the record from the earliest stages of the matter.
  • Hours of manual search eliminated—queries that previously required reviewing thousands of documents across multiple platforms now resolve in a single question.
  • Healthcare-specific precision reduced false positives and ensured the most relevant billing records, clinical notes, and communications surfaced first.
  • Data security exceeded federal investigation standards, protecting patient records and privileged attorney communications throughout the matter.
  • The embedded model meant GWA never operated alone—DecoverAI's team was available to refine, expand, and support as the matter evolved.

Obtaining the correct information on queries that used to take several hours of reviewing and analyzing 1,000s of documents on various e-discovery platforms is now possible in a single question.

Partner, White Collar Crime Practice Group
GWA PLLC

More case studies

View all →

Ready to process millions of documents with confidence?

See how DecoverAI handles the most demanding white collar investigations.

Book a Demo →
White Collar Crime
Download this case study
Enter your details to get the PDF — no demo required.
No spam. Unsubscribe anytime.