Most business owners aren’t skeptical about AI anymore. They’ve read the articles, sat through the demos, maybe even signed up for a tool or two. What they can’t figure out is why none of it seems to be working the way they were told it would.

That disconnect between understanding that AI matters and actually making it produce results inside a real business is the most consistent problem showing up across industries right now. It’s not a technology problem. It’s an execution problem. And it has less to do with the tools a business chooses than with what’s happening before, and after, anyone touches the technology.

The Human Problem Nobody Wants to Say Out Loud

Henry Hays has spent years working with companies across banking, healthcare, energy, construction, and high-growth businesses on AI adoption and implementation. When asked what the most consistent barrier is across all of those engagements, his answer is immediate: humans.

Not bad technology. Not insufficient budgets. Not even poor data, though that comes up too. The biggest thing holding businesses back from making AI work is that the people inside them, including the leaders, aren’t ready to change as fast as the technology is moving.

That sounds harsh, but it’s actually clarifying. If the problem were the tools, the solution would be simple: buy better tools. But if the problem is human behavior, the solution requires something harder: honest self-assessment, leadership that’s willing to model new behaviors, and a company culture that can move without waiting for the perfect plan.

The pattern Hays describes is familiar to anyone who’s watched a new technology roll through an organization. Every industry thinks it’s uniquely resistant to change. Every executive believes their situation is more complicated than the last. Every team has a reason why now isn’t quite the right time. And so the technology sits, underused, while the company watches from the back of the line.

That approach worked fine with slower-moving technology cycles. It doesn’t work with AI, where the distance between early movers and late adopters is compressing faster than anyone expected.

Why Strategy Alone Doesn’t Get You There

There’s no shortage of AI strategy conversations happening right now. LinkedIn is full of them. Conferences are built around them. Consultants are selling them. But strategy and execution are two different things, and most businesses are stuck somewhere in between.

The strategy conversation usually breaks down in the same place: when the responsibility lands on someone who doesn’t actually know how to move it forward. CTOs and CIOs are getting handed AI mandates without the training or frameworks to execute them. So they hedge. They cite data concerns, security risks, compliance questions — not because those aren’t real considerations, but because they provide cover for not knowing what to do next.

What’s missing in most cases isn’t a better strategy document. It’s a practical starting point. A real assessment of where the business actually is: what technology it’s running, whether the data is in any shape to be useful, and whether the people involved have any genuine readiness for what’s being asked of them.

If everything else is ready and the data isn’t, Hays puts it plainly: you’re burning money. A clean, honest assessment before any investment is made is the single step most businesses skip, and it’s the step that determines whether everything that follows works or doesn’t.

Start With the Boring Stuff

One of the most useful reframes for any business owner trying to figure out where to begin is this: stop looking for the sizzle.

The AI use cases that get the most attention, the ones that show up in demos and keynotes, are almost never the right starting point for a growing business. The right starting point is whatever is boring, mundane, and tedious in your team’s Monday-through-Friday workweek.

Think about the tasks your best people spend time on that require almost no judgment. The insurance underwriter who reads a 101-page policy document just to pull out the key data points. The salesperson who manually updates CRM records after every call. The operations manager who pulls the same report every week and reformats it for the executive team. Those are the tasks where AI delivers fast, measurable, meaningful returns, and where a team can start building the confidence and momentum to go further.

The businesses that get stuck are usually the ones chasing a bigger vision before they’ve validated the basics. They want to build a custom AI agent before they’ve automated a single repetitive workflow. They want enterprise-wide transformation before anyone on the team has used AI enough to know what they’re doing.

Start smaller than feels right. The wins will come faster, and they’ll build the internal credibility needed to go bigger.

The Three Boxes Every AI Investment Should Check

When there are competing ideas for where to use AI, and there usually are, the question isn’t which idea is most exciting. The question is which ideas actually pass a practical business test.

Three criteria worth applying to every proposed AI use case: Does it make the business more efficient, in a way that can be measured? Does it make the people doing the work more productive, in a way that can also be measured? And if the first two are true, does it create a path to more revenue?

Any use case that can’t answer yes to all three is probably a distraction. The bar isn’t high. Most legitimate operational improvements will clear it easily. But the filter is useful precisely because it cuts through the enthusiasm and forces a conversation about what the business is actually trying to accomplish.

It also reframes the AI investment decision away from “what’s the most impressive thing we could do” and toward “what’s the most useful thing we should do first.” Those are very different questions, and the second one tends to produce much better results.

What the Leaders Who Get This Right Do Differently

The businesses generating real ROI from AI share something in common that has nothing to do with their technology stack: their leaders are users.

Not users in the sense that they’ve opened a ChatGPT account once or had their assistant set something up. Users in the sense that they’re in the tools every day, running real work through them, developing an actual feel for what AI can and can’t do. They’re not waiting for the technology to prove itself before they engage with it. They’re engaging with it so they can lead their teams through what comes next.

This matters more than it might seem. When a team member hits a wall with an AI tool and brings a question to their manager, the manager who has been using the tool can answer it. The manager who hasn’t been using it can’t, and that moment of uncertainty ripples outward. Teams read their leaders. If the leader isn’t taking AI seriously enough to use it themselves, the team won’t either, regardless of what the official policy says.

The other thing the leaders who get this right tend to do is resist the pull toward over-planning. Analysis paralysis is real in AI adoption, and it usually shows up as a long list of competing use cases that never gets prioritized. The businesses that move pick something small, test it, learn from it, and move again. They’re not reckless. They’ve done the assessment, they’ve set the guardrails. But they’re not waiting for certainty that isn’t coming.

It’s Not Too Late, But the Clock Is Moving

One of the most common questions from business owners right now is some version of: are we already too far behind?

The honest answer is no. But the window for catching up without real cost is narrowing. The businesses that move in the next year will have a meaningful head start on the ones that wait another year after that. And the ones that wait another year after that will be competing against organizations that have two or three years of institutional learning and operational improvement already built in.

Getting started doesn’t require a large investment, a technical team, or a comprehensive strategy. It requires a leader who is willing to become a practitioner, a team that has some basic guidelines about how to use AI responsibly, and a first use case that’s specific enough to actually measure.

The Model T topped out at 36 miles an hour. The people who saw it and thought they understood where it was going were still thinking in horse-and-buggy terms, because that was the only frame of reference they had. The businesses treating AI as a better email tool are making the same mistake. The ceiling is much higher than most people realize, and the ones who figure that out early are going to have a significant advantage over the ones who figure it out later.

Listen to the full podcast episode here.

Related Resources

• The 90% Problem: What the MIT Study on AI Pilot Failure Actually Means for Your Business: The stat Henry Hays cited in this episode, 90% of AI pilots fail, isn’t just a number. This post breaks down what’s actually behind it and what it means if your business is planning an AI initiative.
• The Boring, Mundane, and Tedious: Where AI Actually Delivers for Growing Businesses: If you’re trying to figure out where to start with AI, this post gives you a practical framework for finding the workflows where AI delivers fast, measurable results without a large investment or technical team.
• How to Evaluate Any AI Investment Before You Spend a Dollar: Before committing budget to any AI initiative, there’s a short list of questions every business owner should be able to answer. This post walks through what a real pre-investment assessment looks like.
• Stop Treating AI Like Google: What Business Owners Get Wrong from Day One: The single most common mistake new AI users make is treating it like a search engine. This post explains the mindset shift that changes how much value you get out of every AI interaction.