The firms making real progress with AI rebuilt how they work, not just which platform they chose
The efficiency case for artificial intelligence is clear. In record time, it can speed up many of the rote tasks that lawyers traditionally performed. When productivity gains are real and the tools are widely available, however, why do some firms compound those gains while others stall after the purchase?
Bruce Chapple, chief executive and managing partner of McMillan LLP, recently sought the answer during his firm’s evaluation of Legora, the AI platform it has since adopted firm-wide. Chapple watched for the signal that would tell him whether to commit. It arrived in an unexpected form: usage during the pilot did not plateau – it accelerated. “They were actually, in a very short period, becoming dependent on it, finding new ways to use it, and getting excited about it,” he said. Litigation and regulatory teams are now using AI to extract specific information across thousands of documents at a time; deal lawyers are applying the same approach to contract-heavy due diligence. Resistance within the firm has largely given way. “Even the most skeptical are now saying, ‘This looks pretty cool, some of the things that can happen here,’” Chapple told me.
Fasken reached a similar destination through an entirely different route. The firm co-developed early agentic workflows with Walter AI beginning in 2024, before Legora acquired Walter and the partnership extended across nearly 1,000 lawyers. Clarke Barnes, managing partner at Fasken, described client sentiment completing a full rotation in less than 24 months. “A number of clients were saying, look, we’re not too sure about AI, don’t use it in my work,” Barnes said. “That shifted to, okay, we understand everybody’s using this, tell us how you’re using it. And now they’re saying, use it and indicate to us how you’re using it and how those efficiencies are bringing to bear better outcomes for us.”
Two firms, one platform, two different paths – and the same conclusion: commitment matters more than tool choice.
Where firms lose the thread
Samuel Puchala, president and general manager of LexisNexis Canada, recently told me that he sees both ends of the spectrum. He described a bimodal split among firms approaching AI: one group tries it superficially, gets poor results, and shelves it; the other gets impressive results and relies on it too much, bypassing the professional judgment still required. “There’s a happy middle ground,” he said, “where there’s a very significant amount of efficiency and proficiency which you can gain from these tools. But you have to keep your professional judgment.” His frame for that middle ground: treat AI “as if you were working with a junior associate.”
The firms that succeed tend to share a profile. They “have internal champions,” Puchala said, “and they sometimes discover use cases and the ways to optimize the product that even we as vendors have not necessarily always thought of in advance.” At the other end are firms that know they can no longer sit on the sidelines but haven’t defined what success looks like. “Piloting is fine,” he said, “but you need to have a very strong vision about what good looks like.”
Lindsay Duprey, president of Array Canada, a litigation support and e-discovery services company, recently made the same point in a different way on a CL Talk podcast episode. “Treating legal technology as a tool versus an operating model” is the industry’s core mistake, she said. “Buying the technology is easy. The real challenge is actually embedding it into workflows and figuring out how that delivers value.” Legora co-founder and CEO Max Junestrand framed phase two of adoption the same way in a CL Talk podcast episode earlier this year: “a top-down process reengineering in the biggest practice areas,” rebuilding them “from the ground up with an AI-first mindset.”
At Fasken, that second phase is visible in headcount decisions. Andrea Alliston, a partner and leader of the firm’s knowledge and practice innovation programs and initiatives, described two concrete steps in a recent CL Talk podcast episode: a secondment program placing lawyers into her team for a year at 50 percent of their time to build practice-specific Legora workflows, and the hiring of dedicated AI engagement lawyers for hands-on, in-person support at the practice level. Training, Barnes added, is “really the key” to governance. The goal, Alliston said, reaches beyond current workflows: “We will start to see more interesting opportunities to deliver services to clients… actually being able to do things for clients that we can’t actually do today because it’s just not cost-benefit with humans doing the work.”
Junestrand was clear that AI isn’t always about speed: “It can actually make it take longer to complete a task – but the quality goes up.” AI can handle more of the initial output, but the gain is not always a faster draft. Sometimes it is better thinking.
The firms gathering in Toronto in October 2026 for the Canadian Legal Summit are not debating whether AI belongs in legal practice. That question has been answered. What they are working through is harder: not which tool to adopt, but how deeply to rebuild around it. That is a process question before it is a technology question, and a leadership decision before it is either.
Puchala, Duprey and Alliston will appear as panellists at the Canadian Legal Summit.
