AI LEADERS
Adoption
Half of firms put faith in legal-specific genAI support — but assessing ROI tops the organisational challenges that accompany it
Three-quarters of firm leaders say they’ve evaluated up to 10 genAI tools since the launch of ChatGPT, while some have seen 50 or more
Join Briefing AI Leaders
Get on the mailing list for our exclusive AI research, hear about dedicated roundtables for senior leaders like you making decisions about AI investment, adoption and planning in your peer firms, and be part of the AI Leaders community.
As firms continue to evaluate the application of generative artificial intelligence to tasks across the firm, our research into those leading in the area finds half now have a domain-specific tool at the end of 2025. That means a solution designed and customised specifically with law firms in mind when it comes to factors such as accuracy thresholds, risk management and IP protection, availability of legal resources such as precedents within workflows, and approach to language and distinctions. They are built by those with a thorough understanding of how lawyers work, looking to streamline processes such as contract review, drafting and discovery effectively but with deep knowledge to draw on at the core. Only 2% say an investment like this doesn’t feature in their plans for AI adoption, with over a quarter (29%) still in the midst of trialling solutions before making a strategic decision.
Claire Stripp, head of knowledge and learning at Browne Jacobson says: “GenAI needs exposure to the terminology, context, work types, common partnership model and regulatory environment within which firms are operating to make the experience more engaging and intuitive, and ultimately to deliver the high-quality results and familiar experience that justify the investment.”
Susan Ford, knowledge director at DAC Beachcroft, says: “it’s very important when conducting legal research that the AI tool considers trusted legal resources which can be verified for accuracy. There are significant legal and regulatory implications in getting this wrong.”
50%
of firms have already bought a domain-specific ‘legal’ genAI solution, say AI leaders
Among many possible challenges arising throughout the pilot process, leaders are most likely to point to pressure to assess the return on the investment. Over four-fifths (82%) suggest that this is ‘very’ or ‘fairly’ challenging. Will the use of genAI for parts of work create efficiencies that can be sustained and scaled, adopted strategically across the firm? And where significant time is saved — once human review and oversight are accounted for — does this necessarily translate to financial improvement? Connected, two-thirds describe proving accuracy as very or fairly challenging in itself. Firms must first manage to define the business impact they expect, then measure how far that vision is reflected in delivery. It’s notable that practically tracking patterns in use and outcomes to emerge are regarded as a lesser challenge than some others.
Almost three-quarters (74%) of leaders also see a significant challenge in integrating their tools with already established workflows and matter systems — key to scaling adoption by embedding new norms that support people to work more productively on an ongoing basis, and perhaps proving value (effort rewarded) at an individual level. Training to consistent standards in how to use the tools effectively — for example, prompting, use of resources and verification — is another important factor. Positively, securing the backing of senior lawyers to work with AI appears to be no problem for a fifth of leaders — two-fifths (41%) report it is only 'fairly' challenging.
Michael Kennedy, head of innovation and legal technology, R&D, at Addleshaw Goddard, says: “We invest significant time in calculating ROI, identifying areas of greatest impact, and developing innovative pricing methodologies, in collaboration with our lawyers, pricing and finance teams. Education is a key component, ensuring efficiency gains translate into tangible value for both the firm and our clients. On the integration front, we adopt a mixed buy-and-build strategy, allowing us to tailor solutions to our specific needs while also collaborating closely with vendors to shape product roadmaps and integration priorities.”
Alastair Mitchell, chief operating officer at Cripps, says: “Over time people will engage with AI to expedite process, but returns won’t be as high as people currently anticipate. I personally think they will lead to short-term apathy in certain areas and delay adoption — it will come but not as fast as some think or would like. We are looking at the benefits of each product, but it’s difficult to identify one product as the ‘go to’ as everyone wants different things from AI.”
Nick Pryor, director of knowledge and innovation at Freeths, says: “Things are moving fast, and the market participants are all figuring out what they want to own and excel at, set against the benefits of integrating with the broader marketplace. I would always favour an open, frictionless, and collaborative ecosystem, but I accept that's sometimes easier said than done.”
Almost half (47%) have worked on between one and five pilots in the last three years — another third (31%) between six and 10 — but a number are clearly managing the evaluation workload of many, many more options in the mix.
Andrew Telling, head of knowledge management at Taylor Wessing, says: “With multiple pilots the key is to plan — prioritise and be strategic — so as not to overload people. To test too much at once is counterproductive. You have to pace yourselves and assume it will keep on coming.”