Almost every legal professional in Asia-Pacific now uses generative AI (GenAI) in some form, but far fewer can show what it’s returning.
AI use among Asia-Pacific legal professionals reached 94% in 2026, according to a LexisNexis survey of 1,715 respondents across 10 regional markets. Only half of Australian law firms measure the value AI delivers, according to Carol Chris of LexisNexis ANZ.
Task-first adoption sets the legal teams getting value apart from everyone else. Firms and in-house legal teams that see returns picked one specific task first, such as contract review, and kept a named lawyer on every output. The rest rolled out a platform and hoped.
AI Use Is High But Measured Value Is Not
Most lawyers across 13 Asian jurisdictions were still holding off on GenAI in 2023, according to a Meritas survey of Asian lawyers. That caution has mostly faded, and LexisNexis now puts AI use across the region at 94%. Only 58% of Asia-Pacific legal professionals work under an AI policy, according to the same survey, and fewer still say their organization tracks what AI delivers.
Adoption surveys disagree because Thomson Reuters measures whole organizations while LexisNexis measures individual professionals. Law firm adoption of GenAI sits at 41% worldwide, according to Thomson Reuters. So one associate trying a chatbot counts as adoption for LexisNexis without turning the whole firm into an adopter for Thomson Reuters.
Firms Sort AI by Task, Not by Hype
Most legal AI disappointment comes from buying the wrong tool for the job. The three categories below solve different problems, so a firm that buys a research tool to mark up a supply agreement often ends up with a pilot nobody renews.
Research and Case Law Tools
Research tools search case law, legislation, and commentary, and they quickly map an issue. Research tools still produce citations to non-existent authority, so every citation needs checking.
Contract Drafting and Review
Contract drafting and review tools work inside the agreement itself. They do the first review pass, propose redlines, benchmark clauses against precedent and market terms, and check drafts against house standards, so juniors and partners are on the same page.
Firmwide Workflow and Knowledge Tools
Firmwide tools sit across matters instead of inside one document. They cover matter management, document automation, and knowledge retrieval across the firm’s past work. They help the whole firm run better, but they won’t tell anyone whether an indemnity cap is off-market.
Contract Work Is Where the Gains Show Up First
A first-pass AI review hands the lawyer a marked-up draft to work from instead of a blank page. The tool proposes redlines, flags clauses that drift from house standards, and flags terms that look off-market or are missing altogether. Spellbook is AI contract drafting and review software that runs the first pass inside Microsoft Word. The lawyer finalizes every document before it leaves the firm, according to published accounts of how law firms use AI with Spellbook.
Transactional work suits AI review better than litigation or research because the answer usually sits inside the document. A tool reviewing an indemnity can check it against the agreement’s own defined terms, a precedent, or the firm’s playbook, so the lawyer can verify each suggestion against the page in front of them. Research tools must look beyond the document for authority, and that’s where fabricated citations creep in.
Documented AI deployments report gains in capacity, with teams taking on more matters without cutting lawyers. Those deployments range from solo practices to the largest US firms, and none of them treat the output as final. The lawyer still weighs commercial risk, decides what to concede in negotiation, and signs off on everything that reaches the client.
What Regulators Across the Region Now Expect
The rules still differ in detail, but regulators and courts keep asking firms for the same three things.
- Accountability stays with the lawyer: Legal professionals remain ultimately responsible for all work products under the first principle of the Ministry of Law Singapore’s GenAI guide. The Solicitors Regulation Authority of England and Wales also holds supervisors accountable for work done through junior staff, regardless of the tools those juniors used.
- Confidentiality depends on the vendor’s terms: The SRA expects client data to stay secure, stay out of model training without authorization, and not be kept longer than necessary, so firms check retention periods, model training, and hosting location before a pilot. Client material stays out of public consumer tools, though confidentiality duties vary by jurisdiction and keep moving.
- Substantial use gets disclosed: Singapore’s guide asks lawyers to consider telling clients when GenAI use is substantial or affects the cost of legal services. The Federal Court of Australia expects practitioners to verify AI-assisted work themselves, and the Hong Kong Judiciary’s guidelines ask judges to stay alert to how court users rely on AI.
Start With One Practice Group and One Document Type
A narrow pilot gives partners evidence from their own matters, and four steps keep it honest.
- Pick bounded, repeatable work: Start with first-pass NDA review or playbook markups in one practice group. First-pass NDA review is predictable enough to measure, so pilot results mean something.
- Put a named lawyer on every output: Record who reviewed each document and when. Supervisors carry responsibility for work done through juniors, so the record gives them an answer when a client asks.
- Read the data terms before the pilot starts: Settle retention, model training, and hosting with the vendor before client material goes in. Fixing terms mid-pilot is much harder than walking away first.
- Measure against a baseline: Track turnaround and realization before and after the pilot. Partners trust their own numbers over any case study, and that comparison earns the wider rollout.
What Separates the Firms Getting Value
Across Asia-Pacific, the adoption question has largely been answered, since most lawyers already use AI. The open questions now center on who signs off on the output, where client data goes, and whether anyone can show what the tools return.
The firms getting value can answer all three plainly, matter by matter, because they started with one task and kept a named lawyer on it. That makes the next client conversation about AI easier, and it leaves them well ahead of firms still waiting for clients to raise it.
