Japan's lawyers build AI skills to stay ahead of automation

As part of the Daily Journal series on how legal professionals are using AI this is the second of three articles based on interviews personally conducted in Tokyo by Howard Miller, past President of the State Bar of California and Daily Journal Contributing Editor, and Marc Miller, Ralph W. Bilby Professor of Law and Dean Emeritus, James E. Rogers University of Arizona College of Law.

The interviews were with Atsushi Okada, Mori Hamada & Matsumoto • Professor Ichiro Kobayashi • Tomohiro Okumura, LegalOn Technologies • Takahiro Homma, Renesas Electronics • and Toshimi Itakura, Sojitz Corporation, all of whom are further identified in the articles. All interviews were recorded, transcribed, and sent for review by participants prior to publication.

Disruption at the Desk

How generative AI is redrawing the map of legal work in Japan--threatening specialized tools, shifting work from outside firms to in-house teams, and forcing a profession to confront an uncomfortable arithmetic about where legal expertise is actually worth investing.

Tomohiro Okumura spent five years as a corporate M&A attorney at Nagashima Ohno Tsunematsu, one of Japan's largest law firms, with stints in New York and Bangkok. In 2020, he left private practice to help build LegalOn Technologies, a legal AI platform now used by more than 8,000 clients globally, across Japan, Singapore and the United States. He has seen legal work from both sides of the value chain, and he is not sentimental about what is coming.

"Some of the prospects said: we have GPT, that's easy enough," he acknowledged, describing conversations with potential clients who had decided that general-purpose AI tools made a specialized contract review platform redundant. He has lost at least five opportunities to that argument and rather than dismiss the threat, LegalOn has responded by doing what general-purpose AI cannot: going deeper into the workflow and broader across the legal operations stack.

The disruption playing out in Japan's legal services market is not yet dramatic--the profession is cautious, the enterprise adoption cycles are long and the regulatory environment slows everything down. But the conversations in Tokyo made the contours of a significant restructuring visible: specialized legal tech under competitive pressure from general AI, in-house counsel absorbing work that previously went to law firms and a profession beginning to confront an uncomfortable question about where the value of legal expertise actually lies.

The Specialist's Dilemma

LegalOn's response to the generative AI threat is instructive. Rather than trying to out-compete tools like ChatGPT on general capability, the company has moved in two directions at once: deeper into the legal workflow and broader across the legal operations stack.

On depth: Okumura's central argument is that a general AI tool handle the prompt-and-answer interaction well, but fall short when it comes to the structured processes that legal operations actually require--intake from business departments, drafting, review, negotiation, execution, contract management--even an AI Assistant and now subsidiary governance. "With LegalOn, in-house legal teams now have everything they need to complete all of their work, all in one place," he said.

On breadth: LegalOn recently acquired a German company specializing in governance solutions--software for monitoring and managing subsidiaries. Okumura describes the company's expanding vision with a Silicon Valley-ready analogy: "Contracts are our books. Amazon started with books."

The push into new markets is, in part, a hedge against demographic reality. Japan's shrinking workforce is driving legal departments to find efficiency anywhere they can. "The population is decreasing in Japan," Okumura said, "but the volume of legal work is the same or increasing. That's why we need to seek productivity." For LegalOn, that pressure is both the problem it was built to solve--and the market force carrying it toward the United States and beyond.

"The population is decreasing in Japan, but the volume of legal work is the same or increasing. That's why we need to seek productivity."  -- Tomohiro Okumura, LegalOn Technologies

The existential pressure on companies like LegalOn reflects a broader dynamic that Renesas Electronics' General Counsel Takahiro Homma described with characteristic directness: "Maybe one year ago, two years ago, Copilot, Gemini,general models couldn't do" what specialized contract review tools did. "Now I don't know, more than half. So if the general model develops, and if we can really customize it ourselves, probably we lose the reason to pay money" for a separate specialized tool.

He did not say this as a criticism of any product. He said it as an observation about where technology is heading--and why in-house legal teams need to track those developments closely.

The Shifting Work

Alongside the disruption to legal tech companies runs a quieter but arguably more consequential shift: work that once automatically went to outside law firms is increasingly being done inside corporate legal departments.

The mechanism is straightforward. Toshimi Itakura, General Manager of the Legal Department at Sojitz Corporation, described a recent example. She needed to send an external letter making a legal claim. She drafted it using authorized AI tool, tweaked the wording herself and sent it to outside counsel for review. "They only made some changes," she said. Previously, she would have written a detailed briefing to the firm, waited for a draft, and paid accordingly.

"I foresee in the future there is less work which goes to the law firm," she said. "This is one way to reduce the cost."

At Mori Hamada & Matsumoto, partner Atsushi Okada has heard the same concern--and pushes back with a distinction that will be familiar to any law firm partner who has thought carefully about the question. "It could happen to more standardized services or services based on knowledge only," he acknowledged. But for a firm that provides value-added services--high-judgment, high-complexity work that depends on experience and creativity--the effect is different. "AI is rather helping our capability." By handling first-draft and more basic work, AI frees lawyers to focus on the analysis and judgment that clients actually need from elite firms.

"Tech fluency is now very important. It is inevitable. It has become one of the core professional skills for lawyers."  -- Atsushi Okada, Mori Hamada & Matsumoto

More than 80 percent of lawyers at Mori Hamada are now using generative AI services--a figure Okada cited with evident satisfaction, noting it exceeds what he hears from many U.S. firms. The firm collaborates with AI startups and global AI companies including Harvey, helping localize Harvey's service for the Japanese market, and has taken equity stakes in some Japanese AI startups. "Tech fluency is becoming a core professional skill for lawyers," Okada said. "Firms where more attorneys have that fluency will be the stronger firms."

The Language and Data Problem

Underlying all of this is a structural challenge that constrains Japan's entire AI ecosystem: the training data problem. Large language models were built overwhelmingly on English-language content. Japanese legal data--court decisions, contracts, practice guides--is far less available in machine-readable form.

"The training datasets in English--the size is so different from Japanese," Okada said. "When I personally use AI services, I often find it more convenient to use English-language outputs." The problem goes beyond language: Japanese contracting style, legal traditions, and even the structure of court opinions differ from common law systems in ways that fine-tuning alone cannot easily address. Court decisions in many practice areas are not publicly available in digital databases. The volume and quality of contract data is a fraction of what is available for English-language legal work.

The result is that most AI tools in Japanese legal practice are built on top of global large language models, with Japanese-specific datasets and fine-tuning layered over them. Purely Japan-native LLMs are rare. This creates a dependence on foreign AI infrastructure that sits uneasily with Japan's regulatory caution and its ambitions for technological sovereignty.

For in-house teams at globally active companies, the language gap has a practical dimension as well. Homma's legal team at Renesas handles patent litigation in the United States, customer and supply agreements with U.S. and European counterparties, and M&A transactions that require fluency in both English and Japanese legal frameworks. The AI tools that work best for cross-border work are the English-language ones--which is where the training data is richest.

Japan's legal AI market is, in other words, simultaneously experiencing the same disruptions as every other advanced legal market--the pressure on specialists from generalists, the shift of work toward in-house, the recalibration of what law firms actually need to be--and doing so with a dataset disadvantage that will not be easy to close. How the profession adapts to both at once will shape Japanese legal practice for years to come.