A&O Shearman has approved a five-year artificial intelligence strategy valued at $1.1bn, most of it in partner time rather than cash, under which each practice will devote 3 per cent of its billable hours to developing AI tools.
Partners agreed the plan this autumn. It amounts to more than 5 per cent of the firm’s annual revenue when the value of that time is counted.
What the firm is actually spending
The headline figure is an accounting construct. Rather than writing cheques to vendors, the firm is committing the hours of its most expensive people to building and testing systems that it hopes will change how legal work is done.
That approach has advantages. Lawyers understand the work well enough to specify what the tools should do, and partner involvement gives the project credibility inside a partnership culture where initiatives led by IT departments often stall.
The disadvantage is opportunity cost. Every hour spent on development is an hour not billed to a client, and in a partnership the people bearing that cost are the same people who vote on the strategy.
Why law firms are investing
Legal work is unusually exposed to automation because so much of it consists of reading, comparing and drafting documents. Large language models are good at exactly those tasks, subject to supervision and to the risk of confident errors.
Clients have noticed. Corporate legal departments have been pushing firms to justify hourly rates when parts of the work can be done faster, and some have begun to bring routine work in-house using their own tools.
The strategic risk for firms is that AI commoditises the junior work that has historically funded the training of senior lawyers. If a task that took a trainee two days takes an hour, the pyramid that supports the partnership becomes harder to justify.
The merger context
A&O Shearman was formed in 2024 by the merger of Allen & Overy and Shearman & Sterling, a combination intended to create a firm with strength in both London and New York. Mergers of that kind require integration of systems, cultures and client relationships, all of which compete for management attention with a technology programme.
The AI strategy is also a way of giving the merged firm a common project. Partners from two legacy firms working on the same tools have a reason to align their practices.
The risks
The first is that the tools do not deliver. Legal AI has improved quickly, but accuracy on complex questions remains a problem, and the cost of checking machine output can exceed the savings.
The second is confidentiality. Law firms hold privileged client information, and using external models requires guarantees about where data is stored and how it is used. Several firms have built private instances of models to manage that risk, at high cost.
The third is cultural. Partners are rewarded for billable hours, and a strategy that depends on them giving up those hours requires a change in how performance is measured. Firms that have made progress have typically adjusted their compensation systems to reward contribution to firm-wide projects.
What to watch
The test will be whether the firm’s revenue per lawyer rises as the tools are adopted, or whether the savings are passed to clients in the form of lower fees. In a competitive market, productivity gains tend to become price reductions, which is good for clients and difficult for firms that have borrowed to invest.
A&O Shearman is making a large bet that the gains will accrue to the firms that build the tools rather than to those that buy them. The next few years will show whether that bet is right.
