Your law firm uses AI. What can you prove it has changed?
9 November 2026
On 5 November 2026, A&O Shearman announced a $1.1bn, five-year commitment to AI. The firm says most of that investment will come from each practice devoting at least 3% of its time to building specialist AI tools and services.
For anyone responsible for law firm AI marketing, the headline figure is the least interesting part. What stands out is how the announcement connects the investment to adoption, client uptake, a competitive win and the outcomes the firm says it will be judged against.
That matters, because law firms are entering a different stage of AI communications, and messages that sounded progressive two years ago are losing their value.
AI adoption is becoming the starting point
On the same day, Fieldfisher announced the firm-wide rollout of Legora to around 1,400 lawyers and business services professionals, following a six-week pilot with 137 users. The announcement says relatively little about the software. It focuses instead on practice-specific uses, role-based training, turning the firm’s know-how into shared workflows, governance through its AI hub and AI Board, and client-facing legal products. Dr Jan Wildhirth, Partner and Managing Director of Fieldfisher X, describes the aim as gaining “better evidence of where AI creates value”.
The two firms are investing at very different scales, yet neither is content to say “we use AI”. Both are trying to explain what it’s for.
That’s partly because the basic claim has become ordinary. Many firms can now say they’ve run a pilot, licensed a platform or given staff access to generative AI tools. When competitors all say the same thing, the statement stops helping a client choose between them.
"We use AI" doesn't answer a client's question
A general counsel or finance director is unlikely to be persuaded by a technology rollout in isolation. What they want to know is whether it will improve the service they receive: whether advice will be more accurate, responsive or consistent, more commercially useful or better value.
That doesn’t mean every client wants the cheapest or fastest legal service. Many clients will still pay a premium for judgement, for a lawyer who understands their business and for advice they can rely on when the stakes are high. A good AI story should reinforce those qualities rather than suggest they’ve been handed to a machine.
Generic claims carry a quieter risk too. Phrases such as “AI-powered” or “at the forefront of innovation” invite a sceptical buyer to ask follow-up questions. If the partner in the room can’t answer them, the claim does more harm than saying nothing.
What credible law firm AI marketing looks like
We’d recommend building any law firm innovation story around five questions:
- The problem: what was slow, inconsistent, costly or difficult before?
- The expertise: what specialist knowledge did the lawyers contribute to the solution?
- The change: how is the work now carried out differently, in practice?
- The safeguards: what governance, checking and human oversight apply, and who is accountable for the advice?
- The evidence: what has improved, and how does the firm know?
The expertise point deserves emphasis. Two firms can license exactly the same platform. What distinguishes them is the knowledge, precedents and judgement their lawyers build into how it’s used, and that’s the part clients are actually buying.
Evidence doesn’t have to be a headline statistic. Depending on the project, it might be adoption rates within a practice, time saved on a clearly defined task, less rework, more consistent outputs, or client feedback used with permission. It might be a fixed-fee service that’s now commercially viable, or one the firm couldn’t offer before.
ContractMatrix is A&O Shearman’s generative AI tool for drafting, reviewing and negotiating contracts. Developed with Microsoft and Harvey, it is used within the firm and licensed to clients.
Seen in that context, the firm’s announcement touches most of the five elements. It names the outcomes clients should see: faster access to expertise, less friction, greater cost efficiency and new insights. It states that a qualified lawyer takes responsibility for all advice. And it offers proof points, including a ninefold increase in ContractMatrix client uptake over 12 months, 96% of partners and 93% of lawyers using its AI tools in September 2026, more than 558,000 activities on its tools that month, and a multi-year mandate won after ContractMatrix scored highest in a client’s own blind evaluation against competing options.
These are the firm’s own figures. They haven’t been independently verified, and the announcement doesn’t define “uptake” or give a baseline. But the structure is instructive: each claim is attached to something a client can understand and, in principle, test.
None of this is an argument for publishing confidential information or thin statistics just to have a number. A percentage based on a handful of matters, or a time saving nobody measured properly, will undermine trust with the readers you’re trying to persuade. If the data isn’t strong enough yet, describe the problem and the change honestly and explain what you’re measuring.
The communications problem often begins before marketing is involved
It’s a familiar communications problem. Marketing hears about an AI project once it’s finished and someone wants a news story. By then, nobody recorded how long the task used to take, agreed what success would look like or gathered client feedback. There may also be no permission to mention the client or the matter.
The solution is organisational rather than creative. Involve marketing at the start, and agree what will be measured before a pilot begins. Innovation and knowledge teams can define the baseline. Finance can help assess commercial impact. Client partners can identify where feedback might be sought and whether a client would be willing to be named. Risk and compliance can confirm early what can be said publicly.
This needn’t be elaborate. A one-page note setting out the problem, the baseline, the success measures and the communications plan will save a great deal of retrospective digging.
Smaller firms don't need to compete on investment
It’s easy for a regional or mid-sized firm to read about billion-dollar commitments and conclude that AI is a contest it can’t enter. From a marketing perspective, that’s the wrong conclusion. Clients of a 60-lawyer firm aren’t comparing its technology budget with a global firm’s. They’re asking whether this firm will handle their matter well.
A modest, well-documented improvement can make a more convincing story than a large but vague claim. That might be reducing the time taken to produce an initial document review, making regulatory updates more consistent, or improving how knowledge is shared across a practice. It could be giving clients better access to information, redesigning a fixed-fee service, or improving the consistency of routine client communications.
What matters is relevance. A clear account of one useful change, explained in plain English with appropriate safeguards, tells a prospective client far more about how a firm thinks than a list of the tools it licenses.
There can be practical benefits to size, too. In a smaller firm, shorter lines of communication can make some aspects of evidence gathering and approval easier, such as agreeing a baseline or getting partner sign-off on what can be published.
Proof is what still sets firms apart
Using AI is becoming ordinary. Showing that it has improved something clients value is not, at least not yet.
For law firm AI marketing to be convincing, firms need to build communications into the innovation project from the beginning. That makes it easier to identify the client benefit, gather meaningful evidence and tell the story clearly, whatever the size of the firm’s budget.
Let's talk
If your firm is investing in AI or developing a new way to deliver legal services, Ruche can help you identify the client story and turn it into clear website content, thought leadership, pitch material and client communications.
All content in this article was correct at the time of publication.
