JND Built a Claims Machine by Saying No Before It Said Go Big

A class-action check is a tiny rectangle with a giant backstage. Before the payment appears, somebody has to locate the right people, write a notice a judge will accept and a human might read, build a secure website, answer the phones, reject duplicate or fraudulent claims, calculate awards, move the money and document every step. JND Legal Administration sells that backstage. Its product is the orderly ending to a disorderly legal fight.

The Seattle company sits in a peculiar category: part professional-services firm, part claims processor, part communications agency and part legal-tech operator. It works for plaintiff and defense firms, corporations, courts and government agencies. On a single matter, JND can be asked to behave like a media planner, a database architect, a multilingual contact center, a compliance team, a fraud analyst and a payments operation. The seams between those jobs are the business.

Three initials, decades of scar tissue

JND began in early 2016 with three people whose first initials supplied the name: Jennifer Keough, Neil Zola and David Isaac. They had already worked together in senior roles at a large legal-administration company. Keough had managed complex settlements and remediation programs, first at Perkins Coie and later as an operations executive. Zola moved from securities litigation into claims administration. Isaac had practiced class-action law and then led a major administration business.

This was not a dorm-room disruption story. It was an operator spinout. The founders knew where settlement programs snagged, what judges expected, how counsel behaved under pressure and which bits of infrastructure could not be improvised. Stone Point Capital provided early backing on undisclosed terms. JND says the investor's reputation for allowing management to operate - and not forcing an artificial exit clock - mattered more than the warm glow of a check.

Their experience came with a useful memory. Years earlier, Keough and Zola had helped build a Louisiana processing center for work connected to the BP Deepwater Horizon settlement at startling speed. Zola has described recruiting hundreds of people locally and discovering that commitment to the community could matter more than a perfect résumé. The lesson was not merely that they could move fast. It was that speed works when the operators understand what absolutely cannot break.

“In the first year or so of JND we told a client that we could not handle a sizable project.”Neil Zola, writing about the company's first years

That refusal is the most revealing decision in JND's history. A young company is supposed to say yes, then find the people, servers and coffee. JND declined because a failure on a prominent matter would spend the one asset it could not replenish quickly: credibility. The founders later argued that taking work over their heads could have damaged their standing before they had one. They chose a delayed invoice over a public stumble.

What actually happens after the gavel

Class-action administration starts before a settlement is final. JND can advise on likely class size, records, notice channels, forms, costs and distribution mechanics. After preliminary approval, the job expands: cleanse contact data, send mail and email, place digital or traditional advertising, run a case website, receive claims, staff toll-free lines, track exclusions and objections, report results to counsel and the court, then calculate and deliver approved benefits.

Mass torts add medical records, plaintiff fact sheets and lien resolution. Government redress programs add procurement, accessibility and public-sector security. Incident response can add forensic identification, breach notices and identity or dark-web monitoring. eDiscovery runs on a neighboring track: collect data defensibly, process it, host it, review it with lawyers and machines, then produce the responsive material.

The customer rarely buys a fixed software seat. JND's model is a custom proposal tied to the matter's volume, complexity, communication mix, review burden, hosting needs and distribution method. Standard prices are not public. Neither were the price of Stone Point's investment nor Sedgwick's later acquisition. The visible cost is operational: experienced people, secure systems, temporary capacity and enough redundancy to survive a spike in claimants. This is software-enabled service, not a frictionless app.

The first thing to fail was culture

JND completed four acquisitions in its first five years. The founders' later account is frank: the results were mixed. One target was run by a husband-and-wife team. Near closing, JND learned that two employees were the owners' daughters; after closing came cousins, boyfriends and more. Then, at a lunch with the staff, someone delivered the sentence every integration team dreads: “We don't like change here.”

The spreadsheet had arrived before the anthropology. JND concluded that it should have walked away when the family web emerged, or at least interviewed key employees privately before signing. The failure was not the acquired company's client list. It was the assumption that a service business could be separated from the habits and incentives of the people doing the service. What changed the founders' minds was lived integration: customer-service standards and appetite for change were not soft variables. They were the operating system.

The counterexample was Alloy Services, a small eDiscovery firm whose leaders welcomed change and wanted to grow. Alloy became the foundation of JND eDiscovery. The division built three patented applications - LayerCake, MachOne and OneSearch - and committed early to RelativityOne's cloud platform. JND now describes itself as the only Relativity Platinum partner focused exclusively on RelativityOne, serving law firms, corporations and government teams from forensic collection through production.

By December 2021, JND employed more than 250 people across Seattle, Los Angeles, Minneapolis, New York and Washington, D.C. Sedgwick acquired the company for an undisclosed sum, kept the brand and retained the founders in leadership. The rationale was straightforward: Sedgwick already managed claims and risk at scale; JND brought class-action administration, government redress, legal notice, mass tort and eDiscovery expertise.

AI enters through a very practical door

JND's newest products are narrow enough to be credible. NoticeIQ, introduced in 2025, models audience segments from historical engagement data and tests notice content and channels before a campaign launches. The old sequence was publish, measure, adjust. JND's pitch is to rehearse first. That can be useful when a confusing form suppresses legitimate response or a weak channel mix reaches the wrong people. It is still a model of human behavior, not human behavior itself, so the value depends on relevant historical data and careful expert review.

WatchTower attacks the opposite problem: too many claims of the wrong kind. Generative AI makes forged documents and scaled identity fraud cheaper. JND says its protocol combines machine learning, behavioral analytics, document verification and a changing risk score, checking submissions at intake and again through the program. The goal is not merely to catch a fake PDF. It is to find patterns across timing, identity, devices, documents and behavior without blocking legitimate claimants.

These products reveal JND's market position. Epiq, Kroll, Angeion, Verita, Simpluris and other specialists offer alternatives in settlement administration; large service providers compete in eDiscovery. JND's difference is not a single magic algorithm. It is the attempt to bundle court credibility, experienced project leadership, public communications, secure processing, payments and software into one accountable operator. The bundle is valuable when a matter is too public, regulated or large to tolerate a loose chain of vendors.

The best use of AI here is not replacing judgment. It is making the queue of things requiring judgment shorter, stranger and more important.

When the JND playbook works - and when it does not

A founder can copy the architecture without administering a lawsuit. Start with a workflow your team knows from the inside. Sell the service manually. Record where time, errors and anxiety concentrate. Build software around those repeated bottlenecks. Protect the reputation layer by refusing assignments that exceed the real system. When buying growth, diligence the behavior of the people who must carry it.

The playbook works best when the workflow repeats, errors are expensive, trust affects selection and each project improves the next one. It works poorly when customers buy only on price, volume cannot support security and compliance costs, the underlying rules change faster than a product can absorb them, or the service has no reusable data advantage. It also fails when automation becomes an excuse to make access harder for the very people a program is supposed to reach.

JND's record is not a neat parable. Anonymous employee reviews describe uneven experiences with pace, management and advancement, a reminder that a client-service culture can feel different from the production floor. Claimants also judge administrators by waiting time and confusing status updates, not award badges. In this business, every delayed check is a personal product review.

Still, the company has endured because the problem is real. Since 2016, JND says it has processed more than 450 million claims, distributed over $5 billion and won more than 1,100 matters. It reached Relativity Platinum status in 2026, added a government eDiscovery environment, and launched a $35,000 annual scholarship at Seattle University in Keough's name. Zola retired from daily leadership that April and became Founder Emeritus.

The modest insight under all those large numbers is the one JND learned twice. Capability is not a line in a proposal. It is a collection of people, systems and habits that must hold when millions of strangers click “submit.” Sometimes building it means going big. Sometimes it means leaving the big job on the table.

The useful doors