A lone figure standing at a cliff edge looking across a glowing chasm at a luminous structure on the other side
March 28, 2026 · Brandon Dietz · 9 min read · Thesis

Why implementation is king.

Here is what bugs me. Most people know about AI now. They used ChatGPT once. Some paid for it. But none of them changed how their work runs. That gap — from "I know AI exists" to "AI does my work each week" — is the whole game.

I am not making this up. The numbers are wild.

The stat that gets me every time

McKinsey says 78% of firms now use AI. That sounds big. But only 5.5% get real value from it. Value means more than 5% EBIT gain from AI. Five point five percent. Out of seventy-eight.

Quick note. I checked this twice. McKinsey uses EBIT. That means earnings before interest and taxes. No A for amortization. So the bar is low. Firms still can't hit it. That makes it worse.

The rest are in what McKinsey calls "pilot purgatory." They ran a pilot. They had a meeting. They bought a plan. But the day-to-day work has not moved. 62% of firms try AI agents. Only 23% scale them. More than half tried. Less than a quarter won. That is the gap.

And in Canada, it's even more lopsided

Here is where it gets bad for Canada. Statistics Canada found that only 12.2% of Canadian firms used AI in the past year. That is up from 6.1% the year before. So yes, it is doubling each year. But it is still one firm in ten. And that counts ANY use, not deep use.

The split by industry tells a story. Info and culture lead at 35.6%. Pro services sit at 31.7%. Finance hits 30.6%. But trades? Retail? Home services? Construction? Almost zero. That is funny. Those are the jobs where AI saves the most time per dollar. The folks who need it most use it the least.

The Canadian Federation of Independent Business has a report that is even more direct. 92% of small firms in Canada use digital tools. But only 10% have fully integrated them. Ten percent. Most are in half-integrated purgatory. They have the tool. They do not have the workflow.

Here is the good news. Canadian SMEs that use generative AI save 1.08 hours per day. That is five hours a week. Twenty-five hours a month. Three hundred hours a year of time back per person. And 55% of firms that invest in digital tools see returns within two years. They get a 29% boost in output in year one. For every dollar spent on tech, the average Canadian SME sees about 1.6x back.

The returns are real. The adoption is not.

A luminous bridge made of glowing amber and teal threads being woven across a dark chasm
The bridge is built mid-span. You don't have to see the far side to start laying threads.

So why is it so hard to cross?

I have thought about this a lot. The reason is boring. It is not that AI is too hard. It is that implementing anything — AI, software, workflows — is a long messy list of small choices that no one wants to make. Which model. Which API. Which prompt. What happens when it fails. Where do the outputs go. Who reviews them. What happens when the vendor changes pricing next quarter. What happens when your forms change. What happens when you need to hand it off to someone new.

Each one has a fix. But stack them up and it is a full-time job. Most small firms do not have that to spare. The owner is at 110%. The staff are at 110% too. The list gets longer each Monday. The idea they will take three months to set up AI the right way? That is a joke. It will not happen. I say that as a guy who tried it more than once.

The CFIB data backs this up. The top two barriers to AI adoption for Canadian SMEs are "lack of understanding of AI's benefits" (62%) and "lack of in-house resources" (60%). Those are not technical barriers. Those are implementation-time barriers. Somebody has to sit down and do the work. The business does not have the somebody.

That is why most "AI for small business" content is junk. It is pitched at an owner who has time to go learn. Most do not. They need someone who has done the learning. Someone who has made the mistakes. Someone who will sit down with their real systems and wire it up.

Which is kind of what I want to do

So I picked a side. Obsidian AI Labs does not do AI awareness work. I will not walk in a room and tell you what an LLM is. You know. We all know. If you want that, go find a podcast.

What we do is set it up. Two ways, because people want different things.

Option one: show you how. If your team wants to own the infrastructure long-term. And some should. We walk through the setup. We leave videos and docs behind. We make sure at least one person on your staff can maintain and extend the thing when we are gone. The goal with this path is that you do not need us in 6 months. You use us to get over the initial hump. Then you are running.

Option two: do it for you. If you would rather stay focused on your real business. Selling. Serving customers. Running the team. And have somebody else own the plumbing, we can do that too. We build it. We deploy it. We maintain it. We migrate it when the underlying tools change (and they WILL change, trust me). You pay a reasonable monthly fee and you stop thinking about it.

Neither one is better. They answer different questions. A five-person trades shop probably does not want to turn into an AI tooling shop on the side. A pro services firm with a sharp ops person might want to take it on and keep going. Both are fine. The answer depends on what your team wants to be good at. That is a talk, not a pitch.

The thing nobody talks about

There is a second part I keep trying to explain. It is harder to sell than the build part. It is what happens to your brain once the small stuff goes away.

Most of us live with this hidden math at work. One hour of thinking costs two hours of work to test it. Want to try an idea? Now you have to do the work too. Want to try three ideas? That is six hours of work. So people stop having ideas. Each idea costs too much. Thinking gets pricey. You do what worked last time.

When the execution layer stops eating your week, that math breaks. You have more ideas than you can implement. You have to pick. That is uncomfortable because we are not used to the ideas being the bottleneck. You end up thinking about thinking. That sounds pretentious until you experience it and realize it is more fun. More of your time gets spent on the part that is interesting.

Stop with the control. Let AI take over the menial things. Humans rise to thought leadership and innovation. It changes from a knowledge-based economy to an ideas-based economy. You're only limited to the idea that comes to mind.

Quick aside. One thing that scares people about AI is they think it will take their job. I get it. I do cybersecurity by day. I have seen how automation eats roles. The honest take is messier than the happy pitch. AI takes routine work off your plate. It gives back time. What you do with that time is the real question. Small firms have always been capped by how much they can get done. AI lifts the cap. The work that fills the new space looks different. The people doing it might too.

Where we start

Okay, tying it all back. If any of this resonated, and you want to stop being in pilot purgatory and implement something, email me at info@obsidianailabs.ca. No sales call. No obligation. No consulting hours. Tell me what is eating your week and we will figure out if there is something worth building.

Sometimes the answer is yes. That is a clear AI job. Let us do it. Sometimes the answer is no. Fix your spreadsheet. AI is not the right tool yet. Either way, I will tell you straight. I will not sell you a contract you do not need.

Build the thing. Talk is talk.

Sources

Tags: AI implementation Canadian AI statistics SMB automation
← Back to all posts