So…you bought an enterprise license of ChatGPT?
Eighty-eight percent of companies say they’re “using AI.”
Which, in my experience, is a bit like saying you’re “into fitness” because you own a Peloton that’s become more coathanger than bike. Owning the thing and actually using it to change anything are two very different situations. And right now, a lot of companies are standing next to their very expensive equipment wondering why they’re not seeing results.
The answer, it turns out, isn’t complicated. They bought the AI. They just never really got on it.
The “We’re Doing AI” Problem
McKinsey’s 2025 State of AI report had a stat that made me laugh out loud (in a sad way). While nearly 9 in 10 organizations have AI running somewhere, only about a third are actually scaling it across the enterprise.
One third.
The rest are doing what I’ve started calling “decorative AI.” It exists. It’s technically there. It looks impressive in board presentations. But it’s not actually changing how the business operates any more than that Peloton is changing your cardiovascular health while it holds your dry cleaning.
And 74% of organizations are struggling to achieve and scale value from their AI investments. So three out of four companies have spent the money, made the announcements, probably updated their LinkedIn company pages … and are now quietly wondering if maybe they missed a step somewhere.
(Spoiler: they did. Several steps, actually.)
Decorative AI looks impressive in board presentations. It just doesn’t do anything.
Pilot Purgatory: A Love Story
Let me paint you a picture you might recognize.
Leadership gets excited about AI. Budget gets approved. A pilot launches. The pilot goes … fine. Not transformative, not disastrous, just fine. Everyone nods. Someone updates a slide deck. And then … nothing much happens after that.
The pilot doesn’t scale. The people who ran it move on to other things. Six months later, someone suggests another pilot, and the whole beautiful cycle begins again.
I call this pilot purgatory, and honestly, it should have its own zip code at this point. Nearly two-thirds of organizations were still stuck there as of mid-2025. That’s a lot of companies running on the hamster wheel of perpetual experimentation.
The problem isn’t that these companies don’t want to scale AI. It’s that scaling requires decisions that pilots conveniently let you avoid. Pilots test the technology. Scaling changes the organization. And changing the organization means difficult conversations, shifting responsibilities, and occasionally admitting that the way you’ve always done things might not be the way you should keep doing them.
Much easier to just run another pilot.
Meanwhile, on the Other Side of the Gap
While most companies are busy piloting their way to nowhere, a small group has figured something out.
Only 10% of companies are exceeding their profit expectations from AI. But that 10%? They’re not just a little bit ahead. They’re operating differently.
A friend of mine watched this happen in real time. She runs operations for a logistics company, and last year a competitor started using AI agents for routing decisions. “Seemed risky,” she said. “We decided to take a more measured approach.”
Six months later, that competitor had cut delivery delays by 40%. Forty percent. My friend’s company is still “finalizing their AI roadmap.”
The measured approach is starting to look less like wisdom and more like standing still while everyone else figures out how to run.
The gap isn’t about having AI. It’s about whether your AI does anything besides exist.
The Floor Is Rising. Are You Building a Second Story?
The new reality is that as AI becomes more widely adopted, the baseline keeps rising. What felt like a competitive advantage two years ago is now just … expected. The floor everyone’s standing on keeps moving up.
The companies pulling ahead aren’t just trying to stay on that rising floor. They’re building a second story. They’re creating capabilities that go beyond what’s becoming standard, building something new and better that others haven’t figured out yet.
Standing on a rising floor isn’t a strategy. It’s just not falling behind as quickly. Building a second story? That’s how you actually get ahead.
And building that second story requires more than good intentions. It requires structure. If you haven’t already, take a look at my One-Page AI Work Playbook for how to set up the principles, policies, and playbooks that make scaling actually possible.
What the Fast Movers Actually Did
So what’s different about the companies pulling ahead? It’s not that they’re smarter or have bigger budgets (though some do). It’s that they made a few specific choices that most organizations keep putting off.
They got boring before they got fancy.
The most advanced AI adopters spent serious time on deeply unglamorous work: cleaning up their data, fixing their integrations, making sure their systems could actually talk to each other. Without that foundation, sophisticated AI is just a very expensive way to generate confident-sounding nonsense. The companies trying to skip to the exciting stuff are getting outrun by the ones who did their homework first.
They made training non-negotiable.
AT&T invested over a billion dollars in retraining their workforce for AI. A billion. That’s not a lunch-and-learn series. The companies seeing actual returns aren’t just buying tools and hoping for the best. They’re building the organizational muscle to use those tools effectively, and they’re treating that capability as infrastructure, not a nice-to-have.
They stopped using AI as a helper and started using it as an operator.
This is the big shift. Gartner predicts 40% of enterprise applications will have AI agents by 2026, up from less than 5% in 2025. The leaders aren’t asking “how can AI help us do what we already do?” They’re asking “what can AI do that we couldn’t do before?” Customer support teams cutting call times by 25%. Logistics operations reducing delays by 40%. That’s not the same game played slightly better. That’s a different game entirely.
The Three Questions That Tell You Where You Stand
Figuring out which side of the gap you’re on doesn’t require a consultant or a six-month assessment. Just answer these honestly.
Question 1: Where does AI actually live in your organization?
Not where is it being piloted. Not where is it being discussed. Where is it embedded in how work actually gets done, every day, without anyone thinking about it as “the AI initiative”? If the honest answer is “a few people use ChatGPT when they remember to,” you have your answer.
Question 2: Who wakes up accountable for AI outcomes?
Not interested in AI. Not supportive of AI. Accountable. As in, their performance review includes whether AI is creating business value. McKinsey found that high performers are three times more likely to have senior leaders with clear ownership. If you can’t name that person in your org, that’s a problem.
Question 3: What has AI measurably produced this quarter?
Revenue. Cost savings. Time savings. Something with a number attached that you could defend in a conversation. If you don’t have one, you’re not behind on AI technology. You’re behind on AI strategy. And those are very different problems with very different solutions.
Standing on a rising floor isn’t a strategy. Building a second story is.
The Window Is Closing
The gap compounds.
Companies investing in AI infrastructure now aren’t just getting ahead. They’re building capabilities that make their next AI initiative faster and cheaper. They’re training their people. They’re cleaning their data. They’re creating feedback loops that accelerate learning.
The distance between leaders and everyone else isn’t growing linearly. It’s growing exponentially. And at some point, “we’re taking a measured approach” stops being a strategy and starts being an explanation for why you’re not competitive anymore.
A year from now, having AI agents won’t be a differentiator. It’ll be table stakes. The question isn’t whether your organization will adopt AI. It’s whether you’ll do it fast enough to matter.
• • •
If This Felt a Little Too Familiar
Look, if reading this made you want to check how dusty your own AI initiatives have gotten, that’s probably useful information.
I work with leadership teams who are ready to move past the pilot-and-hope approach and actually build something that works. Not another exploration. Not another roadmap that sits in a drawer. A real strategy with real accountability and real outcomes.
If that sounds like a conversation worth having, drop me a note.
Frequently Asked Questions
What is the AI acceleration gap?
The AI acceleration gap is the growing divide between companies that are truly scaling AI across their operations and those stuck in pilots and experiments that never go anywhere. McKinsey found that while 88% of companies use AI somewhere, only a third are scaling it enterprise-wide. This creates a widening performance gap where a small group of companies pull further ahead while most organizations stay stuck in “decorative AI” that looks good in presentations but doesn’t change how the business actually operates.
Why do so many AI pilots fail to scale?
Pilots fail to scale because they let organizations test technology without committing to organizational change. Scaling AI requires decisions about processes, roles, and accountability that pilots conveniently let you avoid. A pilot can succeed in isolation without anyone having to change how they work. Scaling means changing workflows, shifting responsibilities, and sometimes admitting that existing approaches aren’t working. Many companies keep piloting because it feels like progress without requiring the harder work of transformation.
How do I know if my company is falling behind on AI?
Ask three questions: Is AI embedded in daily operations? Is someone senior accountable for AI outcomes? Can you state the measurable business value AI created this quarter? If you struggle with any of these questions, you’re likely on the wrong side of the gap. Companies ahead on AI have moved past pilots into embedded operations, have clear senior ownership (McKinsey found high performers are 3x more likely to have this), and can point to real numbers showing business impact—not just activity metrics like tools purchased or pilots launched.
What’s “decorative AI”?
Decorative AI is AI that exists in your organization but doesn’t actually change how work gets done. It’s there for board presentations, company announcements, and LinkedIn updates, but it’s not embedded in operations. Think of it like owning a gym membership you never use. It technically counts as “having AI” while delivering approximately zero results. The 74% of organizations struggling to achieve value from AI investments often have decorative AI rather than operational AI.
What does “building a second story” mean for AI strategy?
Building a second story means creating AI capabilities that go beyond the rising baseline—not just keeping up with what’s becoming standard. As AI adoption spreads, the floor everyone stands on keeps rising. What was a competitive advantage two years ago is now expected. Companies that only focus on staying current are just standing on a rising floor, which isn’t a strategy. Building a second story means developing capabilities that are new, better, and beyond what others are doing, creating real competitive advantage rather than just avoiding falling behind.
How much should companies invest in AI training?
Leading companies treat AI training as infrastructure, not a one-time event, with investments proportional to how central AI is to their strategy. AT&T invested over $1 billion in workforce retraining for AI. While not every company needs that scale, the principle matters: AI tools without AI-literate people to use them is just expensive software collecting dust. Effective training is ongoing, covers all levels of the organization, and focuses on building real capability rather than checking a compliance box.
What’s the difference between AI as helper versus AI as operator?
Helper AI assists with tasks when you ask it to. Operator AI handles multi-step workflows autonomously, making decisions and taking actions without constant human direction. Gartner predicts 40% of enterprise apps will feature AI agents by 2026, up from less than 5% in 2025. The shift from helper to operator is where competitive advantage is emerging. Helper AI might draft an email when you ask. Operator AI manages your entire customer service workflow, routing issues, generating responses, escalating when needed, and learning from outcomes.
Can smaller companies close the AI gap?
Yes, and smaller companies often have advantages including less legacy infrastructure, faster decision-making, and less bureaucracy. The gap isn’t about company size—it’s about strategic commitment. A 50-person company that embeds AI into operations, establishes clear ownership, and measures real outcomes will outperform a 5,000-person company still running its fifteenth pilot. Smaller organizations can move faster, test and iterate more quickly, and make decisions without enterprise-level approval chains.
What’s the first step to closing the AI gap?
Get brutally honest about where you actually are—not where you think you are or where your presentations say you are. Most companies overestimate their AI maturity because they count tools purchased instead of value created. The first step is a real audit: where is AI truly embedded in operations, who actually owns outcomes, and what measurable impact has it created? That clarity becomes the foundation for everything else. Without it, you risk building a strategy based on assumptions rather than reality.
