Something is happening across boardrooms right now. Leaders are watching competitors talk about AI, reading headlines about productivity gains, and doing what any reasonable executive would do: deciding that their company needs to get on board. Fast.
The problem isn't the ambition. The problem is everything that happens next.
AI adoption is becoming one of the most widely chased initiatives in business. It's also one of the most widely botched. And the reasons why have less to do with the technology itself and everything to do with how organizations are approaching it. Let's get into it.
The checkbox problem
Here's what top-down AI implementation looks like in practice: a CEO hears that AI is the future, doesn't fully understand what that means for their specific business, and delegates it downward. “We need to be doing AI.”
Someone nods. A task gets assigned. Suddenly, AI adoption has become a checkbox exercise rather than a strategic one.
The goal shifts from how do we work smarter, to how do we tick the box and report back that we did the thing? Employees get access to a tool they weren't asked about, weren't trained on, and don't understand why they need. And then leadership wonders why nobody's using it.
Contrast that with a bottom-up approach (which is the one we've seen work first-hand). There's probably someone in your organization right now who wants to work with AI tools or wants to implement process improvements with AI assistance — but they aren’t sure they’re allowed to. Instead of a blanket mandate, a bottom-up strategy identifies and encourages those practitioners to experiment with AI on the specific parts of their jobs where they feel friction. Which means the people doing the work discover what's genuinely useful.
Leadership's role is to sponsor experimentation.
Leadership's role isn't to dictate the platform or the use case. It's to sponsor the experimentation, create space for it, and then pay attention to what bubbles up. Better yet: bring in someone who's already done that experimentation and can show your team what's actually worth pursuing. The efficiency gains follow.
And adoption ensues because the people using the tools chose them.
Don't make IT own this
Here's what tends to happen when a C-suite decides the company needs AI but doesn't want to manage it: the mandate gets delegated to IT.
And honestly? We feel for IT teams. They didn't ask to own this. They got handed a "make it happen" order from above, and now they're responsible for a strategy they weren't given budget or context to develop. That's an unfair position to be in.
But here's the root problem with IT-led AI implementation: IT is wired to find one solution and scale it. Everyone on Mac. Everyone on Microsoft Office. One subscription, managed centrally, minimal variance.
That approach makes complete sense for infrastructure. It makes almost no sense for AI.
Different AI platforms are genuinely good at different things. Some excel at research and synthesis. Others are better for creative work, or code, or client-facing communication. The one-size-fits-all procurement logic that works for hardware doesn't translate to a space where the right tool depends entirely on what someone is trying to do with it.
That's why the starting point shouldn't be a platform. It should be an honest look at the iterative processes in your organization where automation (or optimization) makes the most sense.
When IT picks a single platform for an entire organization without that context, they're not setting people up to succeed. They're just setting up a very expensive shelf where tools go to be ignored.
Microsoft Copilot: the value menu option of AI platforms
Which brings us to Copilot.
If you're wondering which AI platform is most commonly landing in corporate environments right now, it's Microsoft's Copilot.
Not because it's the best option. Because it's the easiest one to implement.
It's already inside the Microsoft 365 suite that most offices are already using. There's no new subscription to pitch, no new vendor to onboard. It's the value menu option of AI: a default add-on that comes bundled with the thing you were already paying for.
Nobody gets fired for buying Microsoft
Look: nobody gets fired for buying Microsoft. That's a real phenomenon in enterprise purchasing, and it explains a lot about Copilot’s ‘adoption’.
But Copilot isn't good.
We've tested it across users and across tasks, and the feedback has been consistently underwhelming. If you look at any serious AI benchmark, Copilot doesn't crack the top 15 models. There are platforms most people have never heard of that outperform it by a significant margin. The tools that get high marks from users (Claude, Perplexity, and others) don't have an inherent delivery mechanism baked into the software your whole company already uses. So they require more effort to adopt, even if they're meaningfully better.
The result? With Copilot, an employee's first real experience with AI — the one that shapes their entire perception of whether this stuff is even useful — is a mediocre one. And once someone decides that AI isn't worth the hassle, winning them back is an uphill battle.
The pattern behind the failure
Pull back and the picture is pretty clear. AI implementation fails when:
- Leadership mandates adoption without understanding what they're adopting, and delegates ownership without providing direction
- IT inherits the project and applies a procurement framework that doesn't fit the problem
- The default platform gets chosen because it's easy to implement, not because it's good
None of this is about bad intentions. Most of the organizations we see making these mistakes are full of smart, well-meaning people who are just applying familiar frameworks to an unfamiliar problem. The issue is that AI doesn't behave like previous software rollouts, and treating it like one is exactly why so many implementations fall flat.
We've built our own approach differently: from the ground up, with practitioners in the lead and leadership in a supporting role. We spent months testing platforms, building internal capability, and learning what actually moves the needle. And we've seen the difference it makes.
Now we're using our experiences to help leaders of other organizations skip the trial-and-error phase entirely.
AI Unlocked: Get ready to get it right
If your organization is starting to think about AI adoption, or quietly wondering why the rollout you already attempted hasn't taken hold, we built something for exactly that moment.
AI Unlocked is a private, full-day executive immersion — which takes place following two weeks where we learn your business before we ever walk in the room.
[Show and Tell] demonstrated real-world AI use cases that moved our group from skeptical to active practitioners in the span of 2 weeks.
Ian – Director of Business Intelligence, VPI
No generic decks. No off-the-shelf content. We share live demonstrations built around your actual workflows, and a prioritized roadmap you can take to your board the following week. All with the goal of providing you with a bespoke AI adoption plan that works for you and your team.
Because checking a box isn't the goal. Implementing a plan that works is.
Learn more about AI Unlocked, and reserve your session now.