As legal AI moves from early experimentation into daily practice, the biggest challenge for law firms may no longer be choosing the right technology. It may be changing how lawyers work.
Joe Marando has seen that shift from both sides.
Before joining Harvey as Legal Innovation Partner, he helped lead AI and legal technology adoption inside a major Canadian law firm. Now, working directly with Harvey’s law firm and legal department clients, he is focused on helping teams turn artificial intelligence from a promising tool into an embedded operating capability.
That work has become especially relevant in Canada. Harvey’s recent expansion into Toronto reflects a market Marando describes as unusually active in legal innovation, with Canadian firms leaning hard into AI and pushing beyond pilot projects. But as he explains, successful adoption depends on more than access to powerful models. It requires partner buy-in, measurement, training, workflow redesign, and a clear understanding that legal AI transformation is as much a human change-management challenge as a technical one.
In a conversation with LegalTech.ca, Marando discussed where AI adoption is succeeding, why Canada is an important market for Harvey, how firms can move past pilots, and what an AI-enabled profession means for billing models, junior lawyers, law schools, and the future of legal service delivery.
You’ve moved from leading AI and legal technology adoption inside a major Canadian law firm to working with Harvey directly. What has that shift taught you about where legal AI adoption is actually succeeding?
JM: The teams that succeed in driving impactful transformation are both scientific and sociological about their approach. Scientific in that they’re rigorous with the data, looking at where the technology can have the most impact, measuring what they try, and letting the results reshape the plan. Sociological in that they understand change management is fundamentally a human behaviour problem. This is where art meets science, and it means knowing your organisation at the level of people and personas: who are the power users you can engage, and who will motivate an entire team to try something new.
The other piece is humility, and the realisation that there’s power in numbers. The strongest teams keep revisiting their own decisions, because everyone is figuring this out as they go. That’s a big part of where we come in, because helping legal teams run this transformation, not just handing over the technology, is central to what Harvey does. The other innovation partners and I carry a host of market insights from working with hundreds of engagements, and because our clients are partners far more than consumers of the product, they get the benefit of all of it: the deep bench and the patterns we’re seeing elsewhere.
Harvey opened a Toronto office as part of its Canadian expansion. What makes Canada, and Toronto in particular, an important market for legal AI right now?
JM: Canada is something of a trailblazer in legal innovation, and that’s the heart of it. I had a strong sense of this before I joined Harvey, and it’s only been confirmed since. Firms like Davies, Gowling, and Torys are at the bleeding edge of this technology. There are legal teams pushing hard on AI all over the world, and I’m grateful to work with them, but Canada punches above its weight, so an office in a jurisdiction leaning this hard into legal technology was an obvious fit.
Proximity matters. Our office on Bay Street is steps from so much of Toronto’s legal community, and the team is regularly meeting clients in person from coast to coast. We have a sizable client base here, so being local is also a requirement to deliver the depth of support we aim for. This was never a one-time transaction, it’s an ongoing partnership, which is why it matters to have legal engineers, product engineers, go-to-market, and an innovation team on the ground. The work that decides whether a legal team succeeds happens after the sale, and we are here for it.
From your experience working with lawyers, what separates firms that move beyond AI experimentation from those that struggle to turn pilots into daily practice?
JM: First, the partners need to be bought in. Then, the legal teams that turn it into daily practice are the ones that build it into their rituals and meeting cadences, and a number of leaders have done this well. It looks like standing agenda items to discuss AI use at every level of the organisation, the practice group or department, and the smaller teams within them, with regular moments to pause and ask “is there a better way I could be doing this?”
Just as important is having multiple points of contact pushing information out to people, so it’s available just-in-time rather than something they have to go hunting for. And when someone does need to seek more out, give them more than one door to knock on, whether that’s someone formally in an innovation role or, informally, a peer they already know who’s a power user. The strugglers run a pilot in a sandbox, declare success, and never build that connective tissue. Access to the tools was never the problem. It’s whether the organisation has done the work to embed them into how people work day to day. That connective tissue is also where Command Center comes in. For the Innovation and Legal Ops leaders who own the AI program, it closes the loop between deployment and proof, showing how Harvey is actually being used across the firm, benchmarked against peers, with an agentic layer to generate the reports leadership is asking for. The firms that succeed aren’t just embedding the tools; they’re building the measurement infrastructure to show that embedding is working.
How should law firms think about building AI capability: is the bigger challenge technology selection, workflow redesign, lawyer training, or change management?
JM: It’s lawyer training and change management, even though all four matter. Evaluating the technology is the easy part, you can set selection criteria and grade products on how well they meet your needs, and state-of-the-art technology is table stakes these days. The bigger challenge by far is getting humans to adjust to new tools. You can redesign a workflow on paper and still watch lawyers revert to old habits. AI capability isn’t something you buy, it’s an operating capability you build, and the bottleneck is almost always the human side, not the software.
That understanding sits at the centre of how we think at Harvey. One of our core philosophies is that this is a change management problem, so we’re frankly obsessed with making the transition as frictionless as possible, for the people going through the change and the leaders driving it. The practice of law is enormously demanding, and a lawyer is at their highest and best when they’re lawyering, not learning a new tool. We see it as our obligation to take that friction away, which is also why Harvey invests in hiring people like me to be a voice of the customer next to our product teams, hearing what legal teams wrestled with historically and what they’re facing today, so we can keep the product responsive and even anticipate needs before they come up. Command Center is a direct expression of that philosophy for the people leading the AI program. It gives Innovation and Legal Ops leaders the visibility to understand where adoption is stalling, which features peers have already rolled out, and what the data says about where to focus next, so the decisions driving change management are grounded in evidence rather than instinct.
Canadian law firms often have to balance innovation with professional responsibility, data security, confidentiality, and client expectations. How do those concerns shape the way firms should deploy tools like Harvey?
JM: These concerns first shape how a legal team selects a tool like Harvey, making sure it meets the criteria they expect: enterprise-grade security, not training on client data, the right certifications, data residency where required, audit logs, and granular access controls. That’s the entry ticket.
The deeper point is in what the law societies are rightly advocating, which is that responsible use ultimately falls on the individual practitioner. So a legal team’s job is to support its people: educate them on their obligations, educate them on the technology, and select technology that has compliance at its foundation and a frictionless UI/UX, so the easiest path for the user is also the one closest to their obligations. Keeping a human in the loop is a good example, something the law societies have called for and most AI policies require. Harvey is built to make that verification as frictionless as possible: every output can be traced back to an authoritative source, so it’s on the end user to follow that trail and confirm the source is reliable, but to do it far quicker than they could manually.
What legal workflows are proving most valuable for AI today (research, drafting, due diligence, contract review, litigation strategy, client collaboration), and where do you think adoption is headed next?
JM: The tasks proving most valuable right now share two traits: high volume and low variability. Anything that involves synthesizing large bodies of written material or transforming content across contexts plays directly to what AI does best. Extracting key obligations from an MSA, comparing side letter provisions against an LP agreement, cross-checking employee statements against documentary evidence. These are repeatable, well-defined tasks where agents can produce consistent, review-ready outputs at scale.
Where it’s headed is up the complexity curve. As models improve, adoption moves from single-task assistance to agents running multi-step workflows end-to-end, with lawyers supervising rather than touching every step. We’re already seeing this with long-horizon agents that can review a data room and produce a first-pass issues list across hundreds of thousands of documents in a single prompt.
But the more important shift is about institutional knowledge, not just capability. The best future for legal is one where every firm converts its unique expertise into its own AI and uses that to deliver exceptional client service. We’re investing heavily in helping firms do exactly that, through tools like Agent Builder. Firms that embed their own knowledge, processes, and judgment into agents don’t just automate work, they scale what makes them distinctive.
The constant throughout all of this is keeping lawyers firmly in the loop. What’s expanding is the scope of what can be delegated.
There is a lot of discussion about AI changing the economics of legal work. How do you see tools like Harvey influencing billing models, leverage, client service, and the role of junior lawyers?
JM: This is a big question with a few moving parts. On the economics, for the longest time legal practitioners have billed clients for units of sweat to produce a deliverable, and tools like Harvey let you do the same work with far fewer units of sweat, so understandably clients are asking how they get to share in that benefit. I don’t think the billable hour is going away, but it is being repackaged. For example, applying fixed fees for some tasks that AI can systematize and preserving the billable hour for the advisory work that drives the relationship. What we’re seeing is the forward-thinking teams working directly with clients to co-design what an AI-augmented practice looks like, being transparent about where the cost impact may or may not land, and keeping an open mind about whether the billing model should change at all. Ultimately clients are paying their lawyers to navigate risk, and that doesn’t always fit neatly into a tidy unit of measurement.
The junior lawyer question is the one I take most seriously. Having benefited from the apprenticeship model myself, the concern is twofold: will juniors be cut out of the loop, and will they be deprived of the reps they need to build a substantive understanding of their practice? My view, and Harvey’s, is that AI doesn’t replace legal work or the people who do it, it augments it, and that’s just as true for juniors, because the same technology that does the work can enhance the learning. We hear all the time how law students, junior lawyers, and seasoned practitioners alike use Harvey to lean into continuous learning and build on the education they got in law school or in practice. It’s also why we partner with law schools, legal educators, firms and legal teams, and their L&D teams, and consult regularly on training, in Canada as much as anywhere.
As someone who has also taught legal innovation, what should law schools and young lawyers be doing now to prepare for an AI-enabled legal profession?
JM: For law schools, the shift is to treat AI as a tool of practice woven through the curriculum, not a standalone elective, with supervision as the core skill to teach: how to verify output, where the tools come up short, and the ethics that come with them. Students may be in the best position of anyone to succeed, as long as they hold onto the mindset that makes them good students and carry it into practice. Every semester throws a new concept at you, that’s how the whole education system is built, and it’s easy to lose that instinct once you’re in practice and focused on refining what you already know. The ones who keep that lifelong-learning habit alive will do best in a profession changing this quickly.
The human skills matter just as much. Build real fluency with the tools, but invest at least as heavily in human connection and judgement. So much of what clients come to a lawyer for is an opinion from a trusted adviser, and knowing how to give that well is hard to grasp in theory and far easier to see in practice. So seek out the practitioners you look up to, the ones who’ve done this for years or decades, and ask to shadow them, watch how they work, and ask how they got where they are. The technology is augmenting how we practise, but those opportunities are still there, especially for the people who go looking for them.