As artificial intelligence becomes more embedded in legal work, the firms best positioned to benefit may not be the ones chasing every new tool, but the ones building the knowledge infrastructure to use those tools responsibly.
At Beitel Tax Law, that work is being led in part by Caroline Elias, the firm’s Knowledge Management and Innovation Lawyer. Her role sits at the intersection of knowledge management and innovation, with a focus on how specialized legal knowledge is captured, organized, reused, and eventually made available through better systems and workflows.
For a boutique tax firm, that foundation matters. Tax law is highly technical, detail-driven, and full of distinctions where a plausible answer can still be wrong. Elias says AI can create real value in areas such as research aggregation, document review, legislative summaries, and routine drafting—but only when supported by strong governance, clean data, and lawyers who continue to apply judgment.
LegalTech.ca spoke with Elias about how Beitel Tax Law is approaching AI adoption, why knowledge management is essential to innovation, and how boutique firms can use agility and specialization to compete in a rapidly changing legal market.
As Knowledge Management and Innovation Lawyer at Beitel Tax Law, how would you describe your role and mandate inside the firm?
CE: My role sits at the intersection of knowledge management and innovation. I’m focused on how knowledge flows through the firm, how we capture what our lawyers know, and how we build systems that make that knowledge accessible, reusable, and easily updated.
I believe in a boots-on-the-ground approach: working directly with lawyers and support staff to understand where the friction is, and then building infrastructure that reflects how people work. That philosophy is built into the mandate itself: my considerations are always both strategic and hands-on. On any given day I might be mapping out a taxonomy for our internal knowledge base, testing a vendor’s platform against our real workflows, or sitting with a lawyer to understand how they find precedents when a deadline is looming. The goal is to build a foundation that makes us sharper, not just more efficient.
What kinds of pain points or inefficiencies are you most focused on solving for lawyers and clients?
CE: The biggest pain point in a specialized firm like ours is capturing knowledge that lives only in people’s heads, in email threads, or in a document buried three folders deep that only one person knows exists. That institutional knowledge is very easy to lose, and my focus is on building systems that capture it before it walks out the door.
We think about this through what I’d call a “two-brain” model: one side is internal-facing knowledge (precedents, know-how, lessons learned from matters) and the other is what we surface for clients (FAQs, graphs to help understand different legal processes). The connection between those two sides is where I’m most focused right now.
How is Beitel Tax Law approaching the adoption of AI and other legal technologies?
CE: We approach AI with optimistic scepticism. We’re not standing still, but we’re also not adopting tools because of the “hype”. We’ve piloted some AI tools and adopted others, and the through-line in every decision has been whether a tool genuinely serves our lawyers’ work without introducing new risk. I would say that we think about risk quite broadly.
The obvious risks are privacy, confidentiality, and governance. But we’re equally attentive to subtler ones. A tool that doesn’t integrate with our existing systems creates its own problem: documents and decisions that live in one platform but not another, knowledge that gets generated but never captured. And then there’s what I’d call the “easier” risk, that is tools that smooth over a process so thoroughly that they quietly erode the habits of rigour that good legal work depends on. Making things faster is valuable. Making lawyers less sharp is not. Those aren’t always easy to distinguish in the moment, which is exactly why we build evaluation into our adoption process rather than treating it as a one-time gate.
Tax law is highly technical and detail-driven. Where do you see AI being genuinely useful in a tax practice, and where do lawyers need to be cautious?
CE: In some ways, working in a technical detail-driven area of law makes no difference when adding an AI tool. AI will be most useful in tax where it’s most useful anywhere: in tasks that are high-volume and pattern-driven. Research aggregation, first-pass document review, summarizing legislative changes, drafting routine communications, those are real time savings that free up lawyers to do the work that actually requires judgment. And like with any area of law, one of our priorities is making sure our internal knowledge is organized, accessible, and well-governed; that foundation is what makes AI adoption meaningful rather than superficial. If your underlying data is a mess, AI just surfaces the mess faster.
Having said that, in tax law, a wrong answer can be more subtle than in other areas of practice: tax is full of situations where a small technical distinction matters enormously and isn’t obvious. AI tools can be confidently wrong in exactly those moments. The risk isn’t that a lawyer won’t notice a hallucination that looks obviously wrong. Rather, the risk it’s that they won’t notice one that looks plausible. That’s why we treat AI outputs as a starting point that requires a trained eye, not a deliverable.
How do you evaluate whether a new AI tool or workflow is ready to be used in a live client-service environment?
CE: We always start with a pain point: what is a task or element in our practice that isn’t working well, or is taking too long? That problem anchors the evaluation and keeps us honest about whether we actually need a new tool, or whether we need something else entirely.
During the pilot, we look at whether the tool performs on the kind of work we need it to do by testing against real scenarios from our practice and not generic demos. We also consider the technical knowledge of our staff, the training the vendor can offer, and how well the tool will integrate with our existing software and workflows. A tool that works in isolation but creates friction everywhere else isn’t a viable solution.
Then we ask whether we have the governance infrastructure to support it, and if we don’t, what we’d need to develop before we could deploy it confidently. That means clear policies on when and how the tool can be used, accountability for reviewing outputs, and a way to capture feedback when something goes wrong. A tool that outpaces your governance isn’t ready, even if the technology itself is solid.
What role does knowledge management play in preparing a firm like Beitel Tax Law for more advanced AI adoption?
CE: At BTL, KM and AI adoption are intrinsically linked. We see KM as the connective tissue at our firm, and so we see it as essential in the deployment of any AI solution. KM is what makes the underlying data legible; if that knowledge isn’t structured, tagged, and accessible, you’re not getting the value you think you are.
What I’ve been building at BTL is a framework that categorizes our knowledge across different types. From quick practical insights our lawyers capture on the fly, to polished precedents and authoritative gold-standard documents. Each type has different governance and accessibility requirements. Getting that architecture right means that when we’re ready to layer in more sophisticated AI tools, we’re not starting from scratch on the data side.
How do you think boutique and specialized firms can use innovation to compete with larger firms?
CE: Boutiques move faster. We can deploy a tool much more quickly because we have a single practice area and fewer lawyers to train. Solutions that need to work across a large firm with many practice groups will inevitably face more complexity and more friction. We’re more agile and in a fast-moving technology environment, that’s a real advantage.
Looking across the legal sector, how do you see innovation evolving over the next few years as AI becomes more embedded in legal work?
CE: I am genuinely optimistic about the future. We’re at the beginning of a new wave of technology and there will no doubt be bumps as the legal sector embraces AI. But there is a real opportunity to find new solutions to problems that have existed for decades in ways and at speeds that we never imagined. Frankly, it’s exciting.