There is a sense of déjà vu in the air. A new technology that will “change everything.” Vendors promising miracles. Leaders feeling the pressure to jump in before everyone else. We’ve seen this movie before. It was called digital transformation, and it ended badly for most of the SMBs and organizations that took the plunge. Today it is back on the marquee, with artificial intelligence in the starring role.
The good news is that we already know the plot. And when you know how a movie ends, you can write it differently.
We’ve seen this movie before
Let’s go back to the big wave of digital transformation. The researchers’ verdict is harsh. According to a benchmark study by Boston Consulting Group, only 30% of digital transformations actually meet their objectives and produce lasting change. In other words, nearly 7 in 10 fall short of what they promised. McKinsey reaches a similar conclusion, with a failure rate hovering around 70%.
Now let’s look at AI through the same lens. The recent numbers tell a familiar story. In Quebec, 12.7% of businesses used AI applications in production over the twelve months preceding the second quarter of 2025, according to the Institut de la statistique du Québec. That is a rise of just 3.3 points in a year. Adoption is real, but slow.
Meanwhile, the Business Development Bank of Canada (BDC) finds that 30% of Canadian SMBs use generative AI, and that those that do report 24% higher productivity. The edge is real for those who do it well. But the flip side is just as clear. Gartner predicts that more than 40% of agentic AI projects will be abandoned by the end of 2027, for lack of clear business value and adequate controls.
Put these signals side by side. Low adoption. A real gain for those who succeed. And a mountain of projects doomed to be abandoned. That is exactly the curve we lived through with digital. We’ve seen this movie before.
The real cause isn’t the tech, it’s the silo
Here is the question that really matters: why so many failures? And the answer may surprise you, because it has nothing to do with technology.
The tools exist. They are powerful, accessible, often already paid for in your licences. That is not the problem. The problem is the same as ten years ago: the tools arrive in silos.
You plug an AI into one corner of customer service. Analytics software somewhere else. An assistant in the office suite. Each piece works on its own, but none of them talk to the others. The tool inherits the existing mess instead of resolving it. And that is where the real waste happens. One of AI’s greatest strengths, its ability to process a huge amount of information at once, stays locked inside each silo. The power is right in front of you, and it is never used.
That is the real gap between organizations that win their AI shift and those that burn out. Not the budget. Not the trendy tool. The ability to connect what they already have, so information flows instead of sleeping in pieces. An SMB that connects its data beats a large company that scatters it.
Magic is no substitute for coherence
There is another factor to name, even if it is uncomfortable. The market is saturated with promises.
When you closely follow the news, the messaging, the ads and the AI service companies, one pattern keeps coming back. Most of them sell magic. A dream that will “fix everything,” “transform everything,” “last.” Very few sell something coherent, actually wired into your operations. And a dream without coherence ends exactly like yesterday’s digital transformations: an abandoned project, a burned budget, a discouraged team.
There is a simple test to protect yourself. In front of a shiny promise, ask one question: wired into what, exactly? Which data? Which documents? Which real processes? If the answer stays vague, you are not looking at a solution. You are looking at a new silo, prettier than the others, but a silo all the same.
Raising this point is not about painting a bleak picture. It is a matter of method. Better to see the problem clearly so you can aim for the right solution. You do not pick the right solution by closing your eyes to the noise. You pick it by looking the problem in the face, then doing the opposite of what the noise proposes.
Connect before you add: the layer that brings structure
If the silo is what makes projects fail, the solution is not one more tool. It is exactly the opposite: connect.
Before acquiring anything, ask yourself the real question: does it connect to what I already have? Your emails, your documents, your client data, your financial operations. The right approach is an intelligent layer that structures and connects what you already own, and that adapts to how you work instead of forcing you to adapt to it.
The right model is the right mix. On one side, a solid, proven foundation that covers what every organization has in common. On the other, the part that makes you unique: your workflow, your rules, your reality. All of it in a single whole that talks to itself. It is neither generic software imposed on you that forces you to bend your operations, nor endless custom development that costs a fortune. It is the balance between the two.
An important word on security. Protecting your data and meeting your obligations is not a step you add at the end. In a sound architecture, both are built into the structure and run through every decision. A layer that connects everything must, by definition, know who has the right to see what. It is a built-in reflex, not an optional module.
The bar to aim for: ask out loud and get a real answer
To make all this concrete, here is the goal that should guide your starting point.
You should be able to ask your central tool and its AI, out loud: what correlations exist across my performance metrics, my data, my documents, my emails and what lives in my tools? And get a real answer, drawn from everything you already own.
This is not a futuristic luxury reserved for the giants. It is the intent and the goal your approach should have from day one. If a solution does not bring you closer to that capability, being able to query all of your information and get a useful answer, then it is probably pulling you away from coherence and locking you into a new silo.
When it is connected and started right, AI’s power finally wakes up. It draws on all your information at once and helps you decide better. That is precisely where the movie gets a different ending.
Where to start
The temptation at this point is to ask for a step-by-step recipe. That is not how it works, and it is precisely the trap that makes projects fail. The right way in is not a to-do list. It is a posture and a few good questions.
Ask yourself these, honestly. Where does my information live today, and how many different places do I have to check to answer a simple question about my organization? What are my current tools, and which ones actually talk to each other? If I wanted to query all my data out loud tomorrow morning, what would stop me? And before I buy the next tool, does it connect, or does it add one more silo?
These questions are worth more than a ready-made plan, because they start from your reality, not from a template. They reveal where you really stand and what your next useful step is.
That is exactly the role I play in my digital execution architecture mandates. I don’t sell magic. I help SMBs and organizations see their situation clearly, connect what they already own and build a foundation where AI finally works for them. If you want to talk it through, write to me through the contact page. And if you would rather take your own measure first, the pre-diagnostic tool gives you, in a few minutes, an honest starting point on your silos and your next step.
We’ve seen this movie before. This time, you get to write the ending.
