AI Is a Multiplier. Your Firm's Weaknesses Are Coming Along for the Ride.

Statistics Canada recently asked Canadian businesses a simple question: do you plan to use AI in the next twelve months? In the third quarter of 2025, two thirds said no. What surprised me wasn't the number. It was the reason. Not cost. Not privacy. Not the technology feeling too new. Most of them said the same thing: AI wasn't relevant to what they do. 😳

I have a version of this conversation constantly, usually with a Managing Partner or a business owner who is very good at what they do. I understand the instinct. Every second post in your feed says AI will change everything, so the natural response is to stop listening.

But the picture is moving fast underneath that. In the twelve months to March 2026, 65.6% of workers in professional, scientific and technical services said they used generative AI tools at work. Not firms. Workers. Roughly two out of three people in the sector are already experimenting, whether or not their firm has a policy on it.

So the real question for firm owners was never whether to look into AI. It's whether that use is being guided on purpose, with a policy, some training, and regular review, or whether it's happening quietly, tool by tool, with nobody in the partnership knowing where client data is actually going.

AI tends to make the good parts of your business better, and the weak parts worse. It multiplies what is already there.

The Owner's Trap in an AI costume

A big part of my coaching work is helping owners spot what I call the Owner's Trap. To control costs or protect quality, an owner keeps pulling more of the business back into their own hands, until they become the bottleneck, even for work they aren't particularly good at.

AI is the most tempting doorway into that trap I've seen in years. It really can produce a decent version of things you currently pay someone else to do: a first-draft blog post, a rough set of visuals, a basic marketing plan. The math looks obvious. Why pay someone $2,000 a month when you can produce something yourself for $20?

That instinct is worth fighting, not following. The goal was never to bring everything in-house because the tools got cheap. The goal is to spend most of your energy on the two or three things your firm does better than anyone else.

Execution work versus judgment work

What AI is really doing to vendor relationships is splitting the work into two kinds.

The first is execution only. You send a brief, they produce a mostly generic output: the agency posting three times a week on your LinkedIn with content that could belong to any firm; the contractor formatting your documents. If that's the relationship, the economics are changing fast, and you already feel it. This is the work worth rethinking. Not because you should now do it yourself, but because it was never what made you different.

The second is judgment, context, and accountability. Your IT provider knows which of your systems will break if something changes. Your lawyer knows what was actually agreed in your last three deals. Your financial planner knows your whole picture. Human advisors can also be wrong, out of date, or generic; that's not unique to AI. The real difference isn't human versus machine; it's whether the work is generic execution or real judgment, and who is accountable when something goes wrong. A vendor can be sued, dropped, or held to a contract. An AI tool cannot exercise professional accountability.

There's a further reason to keep these advisors close, and owners tend to miss it: they're often better placed than you to use AI well inside their own field. Your lawyer prompting AI on a contract knows what to check for. You, doing the same thing in a field that isn't yours, don't know what you don't know. The likely outcome is that your expert, using AI, gives you faster, cheaper and better work than you'd produce trying to do their job. AI doesn't replace them; it just changes your side of the relationship. You walk into meetings better informed, with sharper questions and a working idea.

AI is a multiplier, not a fixer

To be fair, AI can help fix a weak system. It can pull clean data out of a messy document or turn a rambling explanation into a first draft of a proper process. It's not only a mirror. But there's a harder problem than whether it can help: can you tell when it gets something wrong?

If your month-end close is written down, shared across the team, and done the same way every time, checking AI's work is easy: you know what "correct" looks like, so you catch mistakes fast. But if that close only lives in the head of one senior manager who's been there fourteen years, there's no standard to check against. AI will still hand you something confident and correct-sounding. Whether it's actually right depends on a judgment call nobody ever wrote down.

People in my industry like to say "garbage in, garbage out." That undersells the risk. A bad spreadsheet formula usually breaks in a way you notice. AI speeds up the output; it does not speed up your ability to catch mistakes in it. The real risk isn't bad input leading to bad output. It's a bad process leading to output that sounds completely convincing.

This is why I keep pointing owners toward the boring work first. Write down your workflow. Decide what "done" actually looks like. Agree on the questions your firm should always answer the same way. None of it is exciting, and for years it was easy to put off. It isn't anymore. The firms getting real value from these tools aren't the ones with perfect systems; they're the ones improving their systems while using AI where the work can still be checked.

Generic output is a competitive risk

Here's the second trap, and it doesn't feel like one at all: in a world full of AI-generated content, authenticity is becoming the rarest thing you can offer.

Every firm I work with has something that actually wins them work: a way of running engagements, a clear point of view, the exact thing a client says when they explain why they chose you over the firm down the street. That difference is the whole asset. Using AI on default settings wears it away. If your proposals, posts, client emails, and marketing all come from the same tool with the same generic prompt, they start to sound like everyone else's: competent, smooth, and forgettable. At the exact moment your competitors get the same tools you do, sounding the same is the fastest way to lose what set you apart.

The fix isn't to use these tools less. It's to feed them what's actually different about your firm before you ask for anything: your own methodology, your senior partner's real voice, the exact words your best clients use to explain why they stayed. Used that way, AI doesn't flatten what makes you different. It helps you produce more of it, faster, and put it in front of more of the right people.

Your firm needs an AI policy now

Whatever you use AI for, someone in your firm may already have pasted client data into a public tool to save an hour on a memo. A 2024 KPMG Canada survey found that 24% of generative-AI users had entered private company data into public tools, and 19% had entered private financial data. That figure is nearly two years old now, and in a space moving this fast, assume the behaviour is more common, not less. For CPA and professional-services firms, this also raises obligations under applicable privacy law, alongside your existing confidentiality duties. That topic deserves more than a paragraph, and I'll write about it properly soon. For now: if your firm doesn't have a simple, one-page policy on what tools staff can use and what data they can never enter, that's the first thing to fix this week, before anything else in this article.

What AI hasn't changed

You still need to know who you serve best, and why. You still need KPIs someone actually owns. You still need consistency: a firm that changes direction every quarter will just use AI to change direction every quarter, with more conviction each time. And you still need to know exactly what makes you different from the firm next door. That's the one thing no tool, however powerful, can define for you.

The tools are new. The discipline isn't. Firms with clear values, a real vision, sound processes, and genuine accountability will get further and further ahead with AI. Firms without those things will produce more output and make less real progress. AI doesn't remove the need for judgment; it makes it harder to hide when judgment is missing. The firms that benefit most over the next few years won't be the ones using the most AI. They'll be the ones that know exactly what should never be handed off to it.