Peters gives a ground-level account of AI adoption – from deal work to hiring to vendor strategy
The hype around artificial intelligence in law is over. What has replaced it is messier, more consequential, and playing out right now inside Canadian law firms – in the drafting work that junior lawyers are no longer doing, in the strategic decisions firms are making about which legal technology vendors to trust, and in the new skills that managing partners are looking for when they hire. In a wide-ranging conversation with Canadian Lawyer, Matthew Peters, partner and national leader of transformation at McCarthy Tétrault LLP, gives an insider's account of what AI adoption actually looks like on the ground at one of Canada's largest firms. His overall picture is unambiguous: change is no longer theoretical.
"This is no longer something coming; this is now here," Peters says. "You really need to lean into it."
When Canadian Lawyer spoke with Peters for a 2017 cover story on AI in the legal profession, he warned firms not to get seduced by the hype and urged a focus on process improvement first. Eight years on, he says that advice has shifted. "I think that now I'd be more careful about using the hype as an excuse not to move quickly," he says. Lawyers, he notes, are naturally change-resistant – "we have this narrow psych profile that we all fit into and hate the idea of anything changing" – and that instinct to wait is no longer a defensible position.
The pace of change has made that clear. Peters recently received an email from a lawyer at the firm, incredulous at what an AI tool had done during a term sheet review. What would have taken six hours took one. He says he gets an email like that "probably every other day from lawyers at the firm," describing a near-universal sense of amazement at the tools' current effectiveness. The gains aren't limited to speed. Peters also points to the ability to crawl internal data for market-position insights on deal terms, and to run a quick quality check on work product before it leaves the firm, which can be helpful even if it improves the output by an incremental amount.
The response at McCarthy Tétrault has been deliberate. The firm established a policy framework early – human-in-the-loop, security protocols, training requirements – and has since built its approach around workflows rather than tools. With lawyers in Canada now navigating Co-Counsel, Copilot, Legora, and Claude, among others, Peters says coherence matters. "You just have to understand and hang on to the workflow. Our job is to tell you the best tool for the best part, best prompts, best data," he says. That includes prompting partners to ask questions they wouldn't have thought to ask before: How was this work product generated? Were the relevant cases actually loaded? Was the output verified?
That discipline extends to how the firm thinks about building versus buying. Peters is frank that jumping into tools without understanding their technical limits is a fast way to underperformance. His advice to other firms: Ensure someone on the team is technically literate enough to know where the off-the-shelf platforms end and where custom configuration begins. Python scripts, tailored workflows, for example, can push further. "So, in our workflow we're actually going to [use] some, for lack of a better term, custom or differently configured ways to go beyond those limits."
That thinking is visible in MT❯Forge, the firm's recently launched alternative legal service provider division, led by Toronto-based partner Sahil Zaman. Rather than applying AI as a catch-all solution, MT❯Forge is built around repeatable, high-volume transactional work – fund formation, lending transactions, acquisition agreements – where playbooks built from the firm's own expertise are combined with AI tools and contract lawyers to deliver speed, consistency, and cost predictability. Peters is candid that some of that work is replacing what the McCarthy teams currently handle. "We would rather essentially tackle that ourselves and do that first with clients," he says.
The competitive landscape is also shifting in ways that demand attention. Peters is careful not to name vendors, but his point is clear: Law firms need to understand which technology partners are genuinely aligned with them and which are quietly positioning to compete. "We've already heard narratives of some of these larger companies proposing that you can skip the law firms completely," he says. Those vendors are frenemies at best – useful partners today, potential competitors tomorrow – and he urges firms to engage them directly on strategic direction, scrutinize contract terms, and talk regularly to peer firms. That kind of cross-firm collaboration, he says, is already happening at the North American and global level.
Those same ground-level pressures are reshaping what firms expect from new lawyers, and McCarthy Tétrault is acting on that directly. The firm is co-developing a new course with Western University's Faculty of Law – AI-Enabled Corporate Practice: Business Law in Action – launching in the 2027 winter term, that will put students through simulated transactions and AI-assisted drafting exercises before they set foot in a firm. The driver is straightforward: the lower-level drafting work that has traditionally taught junior lawyers their craft is being absorbed by AI, and the apprenticeship model is fraying as a result. "If they've had the opportunity to have a few reps on drafting, they show up actually with some core skills that the traditional student doesn't show up with," Peters says.
Peters makes a concrete prediction on where this all lands. Within three years – not five – the gap between firms that have fully integrated AI and those that haven't will be visible to clients and to the talent market alike. The firms that have done it right will be more human, not less: lawyers focused on judgment and client relationships, freed from administrative drag. "The firms that have … resisted it," he says, "are not going to be able to compete from a value perspective with clients… and quite frankly for talent."
This article is based on an episode of CL Talk, which can also be found here:
The episode can also be found on our CL Talk podcast homepage, which includes links to follow CL Talk on all the major podcast providers.
