The Greater Vancouver Board of Trade published a report on AI adoption in BC earlier this year. I've read a lot of industry reports. This one has some numbers worth paying attention to.
The number that stopped me: 73% of Canadian small and medium-sized businesses have not yet considered using AI. They haven't written it off. They just haven't gotten there. BC sits at 68%, which puts us slightly ahead of the national average. That's not a stat worth celebrating.
I read reports like this for a living now, which is a strange sentence to type after 18 years of pouring concrete and running crews, but the two aren't as far apart as they sound. Both jobs come down to the same question: where is the effort actually going, and is any of it wasted. The GVBOT report answers that question for AI adoption across the province, sector by sector, and the Okanagan doesn't come out looking great.
"Haven't considered it" is doing a lot of work in that number
It would be easier to explain if the 73% had looked at AI and decided it wasn't for them. That's a normal business decision, and plenty of tools deserve exactly that treatment. But that's not what the report found. It found businesses that haven't gotten to the point of forming an opinion at all.
I hear the reasons every week, and they're rarely about the technology. A shop owner is covering for a sick estimator and doesn't have a spare afternoon to explore a new tool. A GC just came out of a rough winter and isn't spending on anything that isn't a truck or a crew. A manufacturer got burned by an ERP rollout six years ago and lumps every new system into the same category. None of that is irrational. It's just what happens when you're running a business with thin margins and no slack in the schedule.
The problem is that "I haven't gotten to it" quietly becomes "I'm not doing it," and nobody ever decides that on purpose. It just accumulates, one busy quarter at a time, until three years have passed and the company down the road is quoting jobs in half the time.
Why construction is dead last
Construction sits at the bottom of BC's AI adoption numbers, below every other sector the report tracked. Not near the bottom. Last. And it's not because construction people are behind the times, whatever the stereotype says. I spent 18 years in the trades before I did this. The people I worked with were sharper about spotting a bad system than most office workers I've met since.
The reasons construction lags are structural, and they're worth naming because they explain what actually needs to change:
- Most trades and construction businesses don't have an IT person, let alone an IT department. Adopting new software means the owner or the office manager figures it out on top of their existing job, or it doesn't happen.
- The work is distributed. Crews are on different sites, using different phones, in and out of cell coverage. Tools built for people sitting at a desk all day don't map cleanly onto that.
- Margins are already thin. A tool that costs money and doesn't pay for itself fast gets cancelled, and that makes owners cautious about trying anything at all.
- A lot of companies were burned once, by software that promised to fix scheduling or estimating and instead became one more system nobody used past month two.
Every one of those is a real constraint, not an excuse. But none of them means AI doesn't work in construction. They mean it has to be introduced differently than it would be at a software company, with less setup, faster payback, and something the field can use without training that eats a week.
Compare that to a professional services firm, where every employee already sits at a laptop all day, IT support is a phone call away, and a new tool can roll out to the whole office in an afternoon. That's not a fair fight, and it's part of why national adoption numbers make construction look worse than it deserves. The industry isn't resistant. It's structured in a way that makes generic software adoption harder, which is exactly why generic software usually isn't the right starting point.
What the gap actually costs
The companies already using AI are saving their employees up to 125 hours a year each, according to the report. That's more than three work weeks per person, returned to the business annually. For a ten-person company, that's over 1,200 hours a year, which shows up as margin, capacity, and the ability to take on more work without adding headcount.
Say a mechanical contractor runs twelve people between the shop and the field. The estimator spends four or five hours a week chasing down which version of a spec is current, because the client emailed a revision three weeks ago and nobody updated the shared folder. The shop foreman re-enters the same job data into two systems because the scheduling software and the invoicing software were never set up to talk to each other. Neither of those shows up as a line item anywhere. Nobody bills for it. It just quietly eats hours every week, and at the end of the year those hours are gone, along with whatever they would have been worth if they'd gone toward a third job instead of paperwork.
That's the part that's easy to miss when you're inside a business day to day. The cost isn't dramatic. It's not one bad decision. It's the accumulated weight of a hundred small frictions that AI is specifically good at removing, and that nobody gets around to fixing because none of them feels urgent on its own.
What closing the gap actually looks like
It doesn't look like a company-wide software rollout with a launch date and a training binder. That's the version that fails, for the same reasons the failed ERP projects failed: too much change at once, too little connection to a specific, felt problem.
The version that works starts smaller than most owners expect. Pick one recurring task that eats real time every week, something specific enough that you could describe exactly what a good outcome looks like. Not "improve communication," but "cut the time it takes to turn a site walk into a written scope." Not "use AI for scheduling," but "stop double-entering the same job into two systems." Fix that one thing. Let the team see it work before you ask them to trust the next one.
That approach is slower to announce and faster to actually deliver something. It also matches how construction businesses run generally: nobody rebuilds the whole shop at once, you fix the bottleneck that's actually costing you money this month, then move to the next one.
The order matters too. Fixing the wrong bottleneck first, one that feels urgent but isn't actually where the hours are going, burns goodwill you'll need for the next attempt. That's usually where a short, honest assessment earns its keep: not to sell a big project, but to find the one or two places where the return is obvious enough that nobody has to take it on faith.
The objections I hear the most
Every conversation I have about this eventually lands on one of a handful of objections, and they're worth answering directly rather than waving off.
"My business is too small for this." Most of the adoption gap is a size gap. Larger companies are two to three times more likely to have adopted AI than smaller ones, and that gap is widening. But that's a statement about who's had the time and staff to figure it out first, not about whether it works at ten employees. Some of the clearest wins I've seen have been at companies smaller than that.
"I don't trust it with client information." That's a reasonable instinct, and it deserves a real answer rather than a dismissal, which is exactly why governance and setup matter before rollout, not after. A tool used carelessly with sensitive data is a real risk. A tool used deliberately, with the right boundaries in place, usually isn't.
"We tried something like this before and it didn't stick." This is the most common one, and usually the software wasn't the problem. It got dropped into the business without anyone owning the rollout, without training that matched how the team actually works, or without a clear reason to keep using it once the novelty wore off. That's a process failure, not proof the technology doesn't work.
"My competitors aren't doing this either, so why bother now." Maybe not yet. But adoption doesn't move evenly, it moves in clusters. One company in a trade figures out a use case that works, mentions it at a supplier event or a bid meeting, and within a year three more are doing some version of it. By the time it's obvious your competitors have adopted AI, they've already had a year or two of the compounding benefit. Waiting for proof from your specific peer group means waiting to go last.
What the adopters are actually doing
Sixty-nine percent of businesses not using AI say they can't identify a clear business case. That matches what I see every week. Most companies I talk to aren't opposed to AI. They just don't know where to start, so the decision quietly makes itself by default.
What's notable is what the companies who have adopted it are actually using it for. Reducing headcount doesn't even crack the top five reasons businesses give for adopting AI. They're using it to move faster on repetitive work: drafting the first version of a proposal, summarizing a long spec document, pulling numbers out of a pile of invoices. They're using it to find information without pulling someone off a job to go dig through a filing cabinet or a shared drive. And they're using it to make better decisions with data they already have sitting around, mostly unused, because nobody had time to look at it before.
Say a manufacturer runs three shifts and keeps a decade of job cost data in a system nobody has looked at closely since it was installed. That data can tell you which jobs actually make money and which ones only look like they do until you account for the rework. Most shops never get around to pulling that apart, not because they don't care, but because it takes a free week nobody has. That's a task AI handles well: not deciding anything on its own, but doing the tedious first pass so a person can spend their attention on the judgment call at the end.
None of that requires an IT department or a six-figure budget. It requires picking one specific, recurring pain point and fixing it, then moving to the next one.
Where this goes from here
The businesses that figure this out over the next couple of years are going to be hard to catch. Not because AI is magic, but because compounding advantages work that way. A company that saves five hours a week on estimating this year reinvests some of that time into winning more work next year, and the gap between them and a competitor who hasn't started only gets wider.
You don't have to bet the business on a single tool or take a vendor's demo video at face value. Treat "we haven't considered it" as a decision worth revisiting on purpose, with a clear-eyed look at where the hours in your business actually go. Most owners already know, roughly, where the friction is. What's usually missing isn't awareness of the problem. It's twenty minutes to sit down and match the problem to a fix that's actually proportionate to the size of the business.
Okanagan businesses don't have a technology problem. They have an awareness problem, and that gap closes the moment someone decides to look.
I don't think most of the 73% are making a considered choice to sit this out. I think they're busy, understandably skeptical after past disappointments, and short on a clear next step. That's a different problem than not caring, and it's a solvable one.
If you're not sure where to start, the free assessment will show you where AI fits in your operation and what to tackle first.