Case File
41 million documents reduced to under 2% for review • Founded 1988 • Employee-owned • 4 petabytes under management • 500+ corporate and law-firm clients •

Company profile / Legal technology

TCDI Turned 41 Million Documents Into a Human-Sized Problem

The legal-tech veteran spent decades turning discovery into software. Its current trick is more interesting: making AI answer to process, people and the evidence.

A court had already rejected the first answer. An international technology manufacturer needed to purge a competitor’s trade secrets from its network, where the material had wandered through email, shared drives and everyday business systems. Another provider had tried keywords. The court found that approach insufficient. By the time TCDI joined the second remediation effort, the client was under enough financial pressure to plan a sale - a sale it could not complete until the data problem was resolved.

This is the sort of mess that explains TCDI better than any software menu. The Greensboro company processed a scoped 13 terabytes of data that expanded to 22 terabytes, or roughly 41 million documents. It used known trade-secret files as seeds, layered similarity analysis and categorization over them, trained a continuous active learning model, brought in the client’s engineers, and sent the narrowed results to its managed-review lawyers. TCDI built remediation libraries in three weeks. The set made available for human review landed around 400,000 to 500,000 documents - less than 2 percent of the starting count.

41Mdocuments processed
3weeks to remediation libraries
<2%left available for human review

The useful failureKeywords found words. The court wanted proof.

What failed first was not technology in the abstract. It was a thin method applied to a thick problem. Keywords could locate literal matches, but trade-secret material changes shape. It appears in diagrams, fragments, paraphrases, attachments and images. It gets copied into new documents and discussed without using the original vocabulary. A defensible cleanup needed more than a search box.

The change of mind was imposed by reality: a court rejected the initial remediation, ongoing litigation costs squeezed the company, and the stalled sale made delay expensive. TCDI’s answer was not to replace keywords with a shiny model. It combined several imperfect signals and made specialists check the result. Image-heavy files received a parallel review. Engineers validated classifications. A gap analysis caught data created while the business kept operating. Custom reports mapped findings back to individual machines and the enterprise network. TCDI then supplied expert testimony through two rounds of depositions.

Documents traveling through a blue digital stream, illustrating trade-secret identification
THE HAYSTACK, NOW WITH SERVERS. Forty-one million files entered; a reviewable fraction came out. Image: TCDI.

Born before the “e”A software shop that grew up inside litigation

Bill Johnson founded Technology Concepts & Design in 1988 with colleagues who could bridge software development and legal services. Discovery then meant rooms of paper, armies of reviewers and databases that felt exotic. TCDI built early systems to make large litigation searchable and manageable. A thick-client case tool from the 1990s evolved into CVLynx around 2002, combining litigation management with electronic discovery long before legal AI became conference décor.

The company today is both vendor and operator. CVOnyx ingests and prepares hundreds of file types, with advertised processing above 100 gigabytes per hour. CVLynx handles large, complex review with threading, near-duplicate detection, clustering and technology-assisted review. CVFox packages processing, early assessment, review and production into a simpler SaaS product for smaller and mid-sized teams. CVUnity pulls status, access, custodian histories and trends into shared dashboards. Communication Intelligence reconstructs chat exports into readable threads. SMART Data applies OCR, transcription, translation and summarization before review begins.

Then come the humans: project managers, forensic specialists, developers, data analysts and licensed lawyers. TCDI’s Military Spouse Managed Review program grew from a practical problem in 2016, when attorney Jennifer Andres faced another military move with her active-duty husband. Portable review work kept legal talent employed across relocations and created a distributed review team for clients. It is culture doing useful operational work.

“We’ve always focused on solving problems for individual clients, not watching trends.”Bill Johnson, founder and CEO

The business of subtractionTCDI gets paid to make the expensive pile smaller

Discovery can account for 25 percent or more of litigation cost, according to TCDI. Every irrelevant duplicate sent to a lawyer is avoidable spend. That makes the company’s core product controlled reduction: collect the right material, process it cleanly, connect conversations, rank likely evidence, and reserve costly attention for the documents that matter.

The model mixes enterprise software with professional services. Clients pay for hosted review platforms, processing, secure data hosting, forensic collections, investigations, managed review, custom development and advisory work. TCDI does not publish a rate card; the shape and cost of an engagement follow data volume, urgency, platform, security and review requirements. The more legible outcome is avoided work. In a separate breach matter, TCDI reduced 1,060,742 documents by 97 percent before review and reported a 60 percent reduction in overall PII-notification project cost.

Who buys

Law firms, corporate legal and compliance teams, insurers, manufacturers, pharmaceutical companies and government agencies.

What they buy

Software, processing, hosting, forensics, investigations, custom workflows, managed review and data advisory.

What hurts

Too much data, tight court deadlines, scattered chat, privacy exposure and methods that must survive scrutiny.

What changes

Millions of files become a smaller, documented set that lawyers can inspect, produce or remediate.

The anti-lock-in positionIts software is not the whole pitch

The obvious alternatives are global providers such as Consilio, Epiq, KLDiscovery and Lighthouse, or a platform-led stack built around Relativity, Reveal or another review system. TCDI’s difference is deliberately untidy: it supports its own CVSuite, RelativityOne and other industry tools, then wraps them in development and services. A client does not have to pretend one platform is right for every matter.

That flexibility was accumulated through acquisitions. JURINNOV added digital forensics and cybersecurity in 2016. Integreon’s eDiscovery division deepened Relativity expertise in 2019. Cicayda brought legal-hold and social-media collection technology in 2020. Aon’s eDiscovery practice added UK reach, RelativityOne, mobile processing and PHI/PII tools in 2022. Private-equity firm Trivest Partners has backed TCDI since 2020, though financial terms were not announced.

Scale now looks substantial without being consumer-famous: TCDI says more than 500 corporations and law firms trust it with four petabytes of data in an on-site Tier 3 data center. Its security program includes annual risk assessments, quarterly access reviews, ongoing training and SOC 2 Type II examinations. In this market, the unglamorous promise - the data stays controlled and the steps stay documented - is often the one procurement cares about.

AI on a short leashThe Tech Lab makes the model audition

TCDI formally launched its Tech Lab in 2023 to test new tools against actual legal work. The lab runs proofs of concept for accuracy, reliability, compatibility and security, then combines the chosen technology with Lean Six Sigma workflows. This is slower than announcing a chatbot and far more appropriate when a bad answer can leak privilege, miss regulated data or collapse under questioning.

Its 2026 partnership with Altorney makes the position concrete. Altorney’s MARC engine performs GenAI-assisted first-pass review behind TCDI’s firewall and can operate without dependence on one large language model. That offers sensitive and regulated clients control over infrastructure while preserving room to choose models. TCDI’s managed-review lawyers validate AI-assisted work, and quality feedback can send a dataset through another pass.

This approach has boundaries. It works when a matter has representative seed material, knowledgeable subject experts, a measurable objective, clean governance and enough volume for reduction to pay off. It is weaker when the dataset is tiny, the relevant concept cannot be defined, source material is missing, or nobody with authority can validate the classifications. Automation also does not remove collection errors upstream. A clever review model cannot inspect evidence that was never preserved.

The part worth copying

  1. Start with the legal decision, not the AI feature.
  2. Use several signals - search, metadata, similarity, active learning and expert context.
  3. Separate machine prioritization from human accountability.
  4. Measure the funnel at each stage and keep the audit trail.
  5. Run a gap analysis when the underlying business keeps creating data.

A patient company in an impatient marketThe crane keeps folding

TCDI is employee-owned, certified minority-owned and says women hold half of its leadership positions. Its public site counts more than 120 employees worldwide, while commercial company data puts the figure closer to 180. Offices now include Greensboro, Purchase and London. The project-management team reports an average of eight years at TCDI and 12 in the legal industry - continuity that matters when a multi-year case outlives the average software implementation.

The old company logo offers an unusually apt metaphor. An origami crane joined the Japanese heritage of one founder with the aviation background of another. A flat sheet becomes a structured object through disciplined folds. That is also TCDI’s work: take unruly information, apply repeatable transformations, and produce something a legal team can hold.

The next version is already visible. In 2025, TCDI won its second consecutive LegalTech Breakthrough award for use of AI in eDiscovery and earned Relativity’s Data Migration competency. In 2026 it expanded UK leadership, opened a London presence, published new case results and put secure GenAI review into its stack. None of this makes discovery simple. It makes the complexity inspectable - which, in a courtroom, is the more valuable promise.

Software meets litigation

TCDI launches in Greensboro and begins turning paper-heavy legal work into data systems.

CVLynx takes shape

The case-management product becomes a combined litigation and eDiscovery platform.

The Tech Lab opens

Emerging AI and software tools get a dedicated test bench before client use.

GenAI moves behind the firewall

The Altorney partnership adds secure, model-agnostic first-pass review.