James M. Lee and the Trouble With Saving Time

Before James M. Lee was selling lawyers on artificial intelligence, he worked in a Yamaha/Kawasaki motorcycle dealership. He was in high school. The service technicians had definite opinions about tools, and one preference stayed with him: they bought Snap-on. The explanation was tactile. The tools felt right in their hands.

Years later, running LegalMation, Lee returned to that memory when explaining his objections to generic AI products for professionals. A mechanic has a job to finish. A litigator has a document to deliver. Neither gets much comfort from being told that the underlying technology is impressive. The useful question arrives when the tool meets the work.

There is something pleasantly unceremonious about this origin for a software philosophy. No revelation in a computer lab. No elaborate prediction about the future. Just people in a service department choosing equipment they trusted, long before a language model could offer an opinion about it.

Lee would go on to earn a bachelor’s degree at the University of Southern California and a J.D. at Stanford Law School. His legal career took him through Morgan Lewis and Quinn Emanuel, then into a firm of his own. But the dealership supplied an enduring test: would an experienced professional reach for this thing again tomorrow?

Forty lawyers against an army

In 2003, Lee co-founded the boutique that became LTL Attorneys. Leaving Quinn Emanuel meant exchanging the infrastructure of an established litigation firm for the pleasures and surprises of building one. He later acknowledged that the departing lawyers discovered they knew less than they had imagined. Founding a business has a reliable way of editing one’s self-assessment.

The firm grew to about 40 lawyers. Across the table, larger opponents could deploy considerably more people. Lee described the imbalance in military terms: the other side brought armies, while his team needed to work like commandos. Technology offered a way to make a smaller group more effective.

This was a working litigator’s problem. Lee tried cases in state and federal courts and represented companies including Walmart, Thomson Reuters, Symantec and VIZIO. He had to think about how legal work was produced as well as whether a particular argument would win. A persuasive brief still needed to arrive on time.

In 2017, he and LTL were included in the National Law Journal’s Elite Boutique Trailblazers recognition. The firm’s announcement described a six-day federal jury trial in which his team defeated trademark claims seeking $30 million. The experience behind his automation project included the distinctly human business of standing up in court and persuading other people.

The paperwork gets a turn

LegalMation developed from that practical search for efficiency. Lee and his co-founders focused on the early work that follows a lawsuit: reading the complaint, identifying allegations, preparing an answer and drafting discovery. These tasks carry consequences, but they also contain recurring structures. A lawyer can recognize the repetition without wanting to spend another afternoon reproducing it.

The early system used IBM Watson technology to help turn complaints into structured information and generate responsive drafts. Its appeal rested on specificity. The output needed to reflect the allegations in the incoming complaint, rather than merely resemble a legal document. A template with a confident manner would leave much of the original job undone.

Walmart and employment law firm Ogletree Deakins were among the early users. Another LegalMation module, SMARTOBJECT, tackled discovery responses. Lee described a company test involving more than 1,500 requests across over 80 discovery sets. The software produced formatted responses with suggested objections in eight minutes, work the company estimated would otherwise occupy roughly 40 hours.

That is a company-reported example, rather than a promise about every case. It nevertheless explains the attraction for a boutique facing a larger opponent: a pile of requests becomes less effective as a weapon when the first pass can be prepared quickly. The lawyer still has a document to examine. The pile has lost some of its bargaining power.

A better memory for the firm

The product also opened another line of inquiry. Once information inside lawsuits could be organized, the company could ask questions about patterns across matters. By 2020, LegalMation was extending into budget analysis and lawyer comparison. Lee described a system capturing 500 distinct types of legal cases, with information coming from court complaints, firms and in-house teams.

For someone who had managed a boutique, the connection was natural. Drafting is one part of running litigation. Choosing counsel and understanding likely costs are others. A document carries information that may be useful well beyond the immediate response it demands. Software that reads the document can potentially make that information easier to compare.

LegalMation’s advisory network reflected the breadth of that problem. Its announced advisers included legal operations leaders Connie Brenton and Tariq Abdullah, alongside knowledge and pricing specialists. The company was placing its work in conversation with people who organized legal departments and purchased legal services, as well as people who practiced law.

Lee’s ambition therefore stretched beyond faster typing. It included making the accumulated experience of legal work more usable. A firm’s past matters can be a resource, provided someone can get at the relevant information without embarking on an archaeological expedition through old files. Every office has its ruins.

Two minutes is only the beginning

By October 2023, LegalMation announced a $15 million Series A led by Aquiline Technology Growth. The financing announcement said customers had already used its platform that year to answer more than 1.1 million discovery requests in over 30 state and federal jurisdictions. Insurance carriers and large enterprises had become its predominant customers.

Those figures describe company activity at that point in time. They also mark the distance from a boutique’s internal frustration to a product used by organizations with recurring volumes of disputes. The attraction for a buyer handling many matters is straightforward: a small improvement repeated often becomes an operational decision.

Lee remained particular about where the product’s value lay. In a 2024 conversation, he put AI at only about 20 percent of LegalMation’s value. Workflows, preparation, analysis and the interface accounted for much of the rest. The estimate challenges the tempting habit of describing an entire product by the model inside it.

Consider the employment-claim response workflow he discussed that year. The system could produce a draft in about two minutes, followed by perhaps another 30 minutes of a lawyer’s work, compared with up to three hours for the earlier process. The distinction matters: generating a draft and completing a professional assignment are different milestones. Lee’s account left room for both.

The dishwasher has entered the argument

Lee has an unexpectedly domestic contribution to discussions of automation: he prefers washing dishes by hand. A 2024 podcast clip paired that preference with his work automating litigation. It is an agreeable complication in the portrait of a technology founder. Someone can want a machine to handle repetitive legal documents and still enjoy doing the washing-up.

The detail helps explain why his motorcycle-shop analogy travels well. People develop preferences through repeated work. Tools enter routines, and routines have reasons, even when those reasons seem eccentric from outside. Persuading someone to adopt new software requires more than demonstrating that a computer can produce words.

His public conversations have repeatedly addressed the limits as well as the uses of large language models. In April 2024, he joined Bridget McCormack and Zach Abramowitz for the AAAi Podcast at Legalweek. Later that year, his discussion with Abramowitz returned explicitly to specialized tools and the expectations of professional users.

These appearances reveal a consistent public argument: useful automation needs a close understanding of its setting. That is a demanding position for a vendor to take, because the details cannot be wished away. The familiar document format, the client’s response strategy and the point where a lawyer checks the output all belong to the product.

“The hard part is that the billable hour resists every efficiency you hand it.”

James M. Lee, 2026

When the hour is also the inventory

For Lee, the deeper obstacle has become the business model surrounding the work. In announcing his departure from LegalMation’s CEO role in 2026, he described hourly billing as an obstacle to making legal work more efficient. Saving time has an awkward relationship with selling time. His years building software had brought that tension into focus.

Cass Gilmore succeeded him as chief executive. Lee said he would remain a board member and involved as co-founder. The change was accompanied by another undertaking: Merrill AI, a litigation business for high-volume defense work. Rick Merrill and Thomas Suh are his fellow founders in the venture, which remains separate from LegalMation.

Lee described the new design as a law firm with an integrated management services organization providing technology, operations and delivery. His stated priorities include pricing around outcomes, measurable improvements in speed and quality, and transparent information that helps legal departments assess what their spending achieves. These are ambitions for the business, rather than established results.

The move brings him back toward the organizational problem that helped start the journey. At LTL, a smaller team needed a better way to compete. At LegalMation, he built tools other teams could use. With Merrill AI, he is helping build a firm around the tools and the incentives together.

The young employee watching mechanics choose their equipment could hardly have known where the analogy would end up. Yet the question has remained remarkably steady through law school, trials, fundraising and a change of chief executive. Give a professional something useful enough to trust. Then build a place where using it makes sense.