From exploration to strategy — mapping the AI journey
Lizzy Duffy, senior director, client engagement, at Thomson Reuters Institute (TRI), speaks to LPM assistant editor Celeste Rivas about the findings of the firm’s 2026 AI in Professional Services Report and discusses adoption, strategy and next steps for AI technologies
In just over three years, generative AI has revolutionised traditional workflows and roles in the legal industry at an unprecedented pace.
According to Thomson Reuters Institute’s 2026 AI in Professional Services Report, 40% of organisations are already using genAI in 2026, up 18 percentage points from a year ago. Among law firms, 41% of businesses have implemented the technology and more than 80% of law firm professionals are now using it at least weekly.
Duffy notes that genAI’s accessibility is a key driver of this rapid rise in adoption — witnessed through three distinct phases: investigation (2023), experimentation (2024) and widespread arrival of genAI in professional services organisations (2025). Unlike other pieces of technology with a longer learning curve, “genAI models are very intuitive and speak your language,” she says.
The middle phase of experimentation has been particularly important, highlights Duffy: “That’s the time when professionals built comfort and confidence in using AI tools. It’s also where firms learnt a lot about what works, which use cases to drive forward and how to use it more effectively. All of that really builds towards a more purpose-driven AI strategy, which is essential for success — as our 2025 research shows, firms with a clear AI strategy are three times more likely to see benefits from AI integration than those without one.”
Shifting attitudes
TRI’s research suggests that professionals’ sentiment towards the future of genAI has shifted: hesitancy has fallen by 12 percentage points between 2024 and 2026, while hopefulness has risen by nine percentage points.
“Since AI arrived in the legal industry, professionals have moved from thinking of AI as a theoretical, futuristic concept to real hands-on use and having a go themselves. So, lawyers are more realistic about what to expect and how to work with it,” says Duffy.
More importantly, this has broadened the perception of what AI’s benefits are beyond time and cost efficiencies, she remarks. Firms are now seeing that it can also help them offer a better and more personalised service to their clients and build more meaningful relationships with them.
She adds: “Leveraging the efficiency and automation that AI offers to gain scale and capacity can give SME firms the ability to complete more complex work for their clients, as well as supporting more predictability around pricing — that’s a real game changer. The firms that are planning strategically now are going to leapfrog ahead of the competition.”
The client gap
Firms are increasingly gaining clarity around AI use cases and strategies, with both clients and law firms looking at harnessing this tech on the practice side of law. Data shows that while 72% of clients believe genAI should be applied in legal work, only 62% of law firms think so.
However, each side has different motivations behind this interest. On the client side, general counsels focus on automation of low-value work to deliver time saving and lower spend on legal services — reflected in cheaper bills, or more work being done in-house.
“Clients’ work is growing in volume, pace and complexity, but their budgets aren’t increasing. They’ve got to find a way to get the work done, and AI is a huge solution for them,” she elaborates.
Meanwhile, law firms expect any passed-on savings to be recouped through offering new or enhanced service lines to their clients as advances in AI enable more systematised, productised and even pre-emptive legal services.
This slight misalignment is one that firms need to address sooner rather than later, says Duffy. “If they are going to make the shift to AI-enabled legal work, firms have to think about what kind of work they want to do, for whom, and which part of the market they want to occupy.
“What I see in the data is that clients aren’t waiting for firms to catch up, so there is a real need for firms to figure this out now or risk being left behind,” she elaborates, noting that whatever strategy they opt for will likely require some operational change. She adds: “The last year has seen a divergence in firms who have taken meaningful steps along this transformation journey and those who are still experimenting.”
To avoid frictions, Duffy advises firms to proactively drive open and honest conversation with clients about AI use to explore different approaches and possibilities — as well as how they can add value beyond the technology. “If you’ve got 10 lawyers having these discussions with major clients to understand their needs, expectations and what guardrails they need to feel comfortable with AI use, they will have crucial and valuable data to shape a consistent firm-wide AI strategy,” she says.
The ROI conundrum
Even as organisations increasingly understand the benefits AI tools can bring in terms of efficiencies, many still struggle to measure this technology’s return on investment (ROI). Currently, only 15% of law firms are tracking it, according to TRI research.
Duffy explains that multiple data points are needed to calculate ROI. This data lives in a range of different IT systems, which can make collating it a complex task, especially in smaller firms.
More importantly, she points out, finding the right success metrics is essential to calculating returns, but painting a more complete picture may be challenging: “A law firm’s strategy has to include AI’s overarching goals and key benefits. If it’s just about improving financial performance, then they’ve probably got the data. But if it’s about driving stronger client relationships and higher client satisfaction, increasing retention and engagement among their talent pool or improving their position in the market, those things are not being measured systematically today at all firms.”
Next stop: agentic AI
GenAI has already started to shift the legal firm landscape. So, what lies ahead?
TRI believes agentic AI could be the next step for innovation. This technology can carry out multi-step workflows autonomously, such as transforming a natural-language query into a legal research plan, executing it and drafting outputs, while still allowing for human intervention at any stage. That gives it the potential to deliver even greater efficiencies than genAI alone.
However, it might take a while for agentic AI to reach the level of widespread adoption genAI has achieved. While the legal industry is already dipping its toes in the agentic AI waters, uptake remains modest: only 16% of the law firms surveyed by TRI say they are currently using the technology.
Duffy believes that there’s a degree of caution, as in the early days of genAI: “I think agentic AI may be a couple of steps out of the comfort zone for many lawyers. The legal work is their expertise and their value — it’s what they bring to the table. And so, it’s high stakes. There’s an accuracy concern that’s fair, but I think there’s also an element of emotional attachment to that work that holds them back from embracing the possibilities of agentic AI. And there is also the need to ensure the right tool is applied to the right job.”
An agentic AI solution can mean many different things, Duffy continues — the real questions are what task the agent is performing, what role the human retains and whether the process preserves accuracy, accountability and contestability. Moreover, it is crucial to understand whether that tech-enabled workflow is a significant enough improvement to justify the investment and operational change. Ultimately, this depends on the nature of the work involved and the outcomes sought.
Against that backdrop, adopting agentic AI requires more than simply adding a new tool; it calls for a deeper rethinking of traditional workflows, responsibilities and safeguards. Duffy sees this as a step forward rather than a barrier — especially because genAI has already laid much of the groundwork.
She notes that firms with more mature AI strategies will find it easier to incorporate agentic technologies effectively and gain a growing competitive advantage. The greater risk, she argues, is having a wait-and-see mindset: “There are still firms who may be waiting to have perfect data or systems and figure out exactly which approach they should take. But they still haven’t started planning for the future in a meaningful way. And I think they’re really at risk of not just being left behind but taken off the map altogether.”
