The bar used to be "do they use AI." That bar is gone. Everyone uses AI now — your accountant, your dentist, the person who just cold-pitched you on LinkedIn. Having a ChatGPT tab open doesn't qualify someone to advise you on how to adopt it inside your business.
The question worth asking is different: do they build with AI? Have they actually assembled something — a workflow, a pipeline, an automated process — that runs in their business without them babysitting it? Not a demo. Not a deck about transformation. Something operational. Something that was doing work while they slept.
There's a real difference between using AI and understanding how workflows, pipelines, and systems work together. The first is table stakes. The second is what you need from someone advising you on adoption. And most of the people selling AI strategy right now have the first and not the second.
This is a buyer's guide. Not for AI tools — for the people selling you AI advice.
The Numbers Don't Mean What You Think They Mean
Statistics Canada says 12.2% of Canadian businesses use AI to produce goods or deliver services — up from 6.1% the year before. That sounds like slow adoption. Meanwhile, 50% of office employees report using AI for work, and other surveys say 66% of small and medium businesses use AI for at least one business function.
So which is it? Are we at 12% or 50% or 66%?
The answer is: all of them, because "use AI" means wildly different things. The 12.2% number is businesses that have integrated AI into core operations — into what they make or sell. The 50% number is employees using ChatGPT on their personal accounts to draft emails or summarise meeting notes because they don't know how else to get through the day. The 66% number is somewhere in between — businesses that have adopted AI for something, even if it's just automating invoices or scheduling social posts.
If you're a small business owner, you're being sold against the 66% number and measured against the 12% one. And the gap between those two is where most of the confusion — and most of the bad advice — lives.
Look Inside Before Anyone Walks Through the Door
Before you talk to a single AI consultant, ask yourself three questions about your own operation:
What really works well, but only because one person makes it work? There's often someone — let's call her Judy — who enters numbers into four different spreadsheets every week to make your reporting accurate. Or someone who manually reconciles orders between your e-commerce platform and your inventory system. Or someone who copy-pastes customer inquiries from three different inboxes into a single tracking sheet. That's not a people problem. That's a process problem that AI might actually solve — but only if someone looks closely enough to see it.
What happens when that person goes on vacation? If the answer is "everything breaks" or "we just wait until they get back," you've found a dependency. And dependencies are where small, practical AI wins often hide.
What takes a lot of time but is highly repetitive? I'm talking about the work where you click here, then click there, then click there again, doing the same sequence over and over. Data entry. File uploads. Status updates. Copying information from one system into another. If you can describe it as "I do this exact thing 40 times a day," there's probably a way to automate some of it — but you have to name it first.
Most AI advisors won't ask these questions. They'll ask what your goals are, or what your pain points are, or what your digital transformation roadmap looks like. Those are fine questions, but they're abstract. The real opportunities are concrete, specific, and usually sitting right in front of you in the form of Judy and her four spreadsheets.
If someone walks into your business and doesn't ask to see how the work actually gets done — doesn't ask to watch Judy work, doesn't ask what breaks when she's not there — they're not looking at your operation. They're looking at their commission.
What to Look For in Someone Selling You AI
Here's what I'd want to see if I were in your position:
Do they build with AI themselves? Not "are they familiar with it," not "do they use ChatGPT" — have they actually built something? Can they walk you through a process they automated, a workflow they put together, something that runs in their own operation? If the answer is vague ("we leverage AI-powered solutions") or the only examples they have are from clients, walk away. You want someone who's done the work on themselves first.
Will they show you? This one's simple. Ask them to show you something they built. Not a vendor demo — something they actually made. A workflow. An integration. A process that runs. If they pivot to a slide deck or a client case study instead, that's a red flag. Real builders can open a laptop and show you the thing.
Do they look at your processes first, or lead with a solution? If the first thing they want to talk about is the platform they recommend, they're not consulting — they're reselling. A good advisor starts by understanding what you actually do, how you do it, and where the friction is. The tool comes after that, not before.
Do they refuse to guarantee ROI? I know that sounds backwards. But here's the truth: results vary. They vary based on whether your team is willing to use the tool. They vary based on whether leadership buys in and rewards people for trying new things. They vary based on how well the tool fits the actual work, not the theoretical work. Anyone who guarantees you'll save X hours or generate Y revenue without knowing those variables is either lying or inexperienced.
Do they talk about what AI can't do? If someone only talks about the upside — the efficiency gains, the time savings, the transformation — they're not being honest. AI is bad at a lot of things. It hallucinates. It requires clean data to work well, and most small businesses don't have clean data. It needs guardrails, especially if you're working with customer information or anything sensitive. If the person selling you AI doesn't name the risks, the limitations, and the work required to use it safely, they're not thinking about your business — they're thinking about closing a deal.
What to Walk Away From
Just as important: here are the things that should make you end the conversation.
"Here's this tool — your problem is solved." No. Tools don't solve problems. People using tools solve problems, and only if the tool fits the work and the people are trained and supported in using it. A ready-made solution that gets dropped into your operation without adaptation is a waste of money.
Guaranteed ROI. I said this above, but it's worth repeating. If someone promises specific results without understanding your team, your data, your processes, and your willingness to change how you work, they're selling you a fantasy.
They can't show you something they built. Talking fluently about AI is easy — there's no shortage of content to absorb. But if they can't point to a workflow they built, a process they automated, or a system they assembled in their own operation, they're theorists. Theorists make expensive advisors.
They don't ask about buy-in or change management. Adopting AI isn't a technical problem. It's a people problem. If no one on your team wants to use the new tool, or if they don't trust it, or if leadership doesn't reward experimentation, the tool will sit unused. A good advisor knows this and asks about it upfront.
The Honest Part: Results Will Vary, and Here's Why
Let me be direct about something most AI consultants won't tell you: even if you find the right advisor, pick the right tool, and implement it well, your results will depend on things outside the technology.
Canada ranks 42nd out of 47 countries on AI trust. Sixty-seven per cent of Canadians say AI makes them nervous. Only twenty-four per cent have received any AI training. That's not a small thing. If your team doesn't trust the tool, they won't use it — or they'll use it badly, which is worse.
And here's the other thing: your team is probably already using AI. They're just doing it quietly, on their personal ChatGPT accounts, because there's no organisational policy and they don't want to ask permission. That's not a problem to solve by banning it. That's a problem to solve by helping them do it well — giving them tools they can trust, teaching them what AI is good at and what it isn't, and building guardrails so they don't accidentally put sensitive information somewhere it shouldn't go.
The question isn't whether your business will adopt AI. The question is whether you're going to help your team do it thoughtfully, or pretend it isn't happening until something breaks.
What Real Use Looks Like
I'll tell you what I use, because I think it's useful to have texture around what "real use" actually means.
I use Wispr Flow. It's a dictation tool that transcribes what I say, cleans it up, and drops it into whatever application I'm working in. That's it. It's a small thing. But it means I can talk through a draft instead of typing it, which is faster for me and easier on my hands. I get more done. I correct myself as I go. At the end of the day, it's not transformational — it's just helpful.
Is that AI? Kind of, yeah. It's processing speech, transcribing it, cleaning it up a bit, and feeding it into the right place. It's not magic. It's automation that makes one part of my work a little bit easier.
That's what most useful AI looks like in a small business: small, specific, and focused on a real repetitive task. Not transformation. Not disruption. Just a tool that does one thing well and saves someone time or effort in a way that compounds over weeks and months.
Where to Start
If you're looking for someone to help you adopt AI thoughtfully — someone who'll do the kind of work I've described here — that's the kind of work I do. I'm not going to sell you a platform. I'm not going to guarantee ROI. But I will look at your operation, ask about Judy and her spreadsheets, and help you figure out where a small, practical win might be hiding.
If that sounds like what you need: [email protected]