If you’ve spent any time around executive teams lately, you’ve probably heard some version of the same question:

How do we make AI work for our business?

It’s a fair question. AI is everywhere. Every week brings another impressive demo, another breakthrough model, or another company claiming transformational results.

But after more than 25 years leading technology organizations, including three tours as an enterprise CIO, I’ve learned something that surprises many executives:

The biggest challenge with AI has very little to do with AI.

It has everything to do with leadership.

I’ve watched organizations invest heavily in promising pilots that never become production systems. In other cases, impressive demonstrations earned standing ovations in the boardroom, only to disappear quietly a few months later. And then there are the companies that get it right: they build an AI strategy that continues delivering measurable business value year after year.

The difference isn’t the technology; the difference is whether someone is truly accountable for the outcome.

That’s one of the central ideas behind my book, Why AI Fails. The organizations that succeed don’t simply buy better technology. They build the leadership discipline required to turn technology into business results.

I’ve Had to Live With the Decisions

One advantage I’ve had throughout my career is that I haven’t simply recommended technology strategies — I’ve had to run them.

As an operating CIO, I didn’t have the luxury of presenting a roadmap and moving on to the next client. Every decision eventually became my responsibility. If something didn’t work six months later, I owned it. If a system failed in production, my team answered the call.

That experience changes how you evaluate technology and AI strategy. When you’ve been responsible for day-two operations, you stop asking whether something can work on the technical level. You start asking whether the organization can realistically support it, govern it, and sustain it over time. Those are very different conversations. That gap is what most advisors get wrong. The recommendation is the easy part. Accountability for what happens next is the hard part, and most advisory engagements end before it starts.

The Hardest Decisions Aren’t About AI Strategy

Many people assume executive discussions revolve around selecting the best model or choosing the right platform. In reality, those conversations usually (or at least should) happen much later.

The first questions should be business questions:

• Who owns the risk?
• What happens if adoption stalls?
• How will this investment be measured?
• What will the board ask six months from now?

Those are the questions that should shape every technology decision, and ultimately determine whether the investment survives.

When I advise CEOs, boards, and private equity sponsors, I spend far more time discussing accountability than algorithms. Technology can usually be changed. Governance decisions are much harder to unwind once an organization begins scaling.

I’ve sat in enough boardrooms to know that every executive carries a limited amount of political capital. Every major technology initiative spends some of that capital. Successful leaders understand this. They don’t propose initiatives that require unlimited organizational patience. Instead, they build confidence by delivering measurable outcomes.

AI Pilots Are Easy

Pilots receive most of the attention. Production deserves most of it.

Let’s face it: almost every AI pilot looks promising. The data is usually clean. The environment is controlled. The users are enthusiastic because they volunteered to participate.

Production is different.

Now you’re dealing with security reviews. Data access becomes more complicated. Integration with existing systems takes longer than expected. Someone has to define a support model. Someone has to answer the phone when an automated process fails in the middle of the night.

Eventually, every organization reaches the same question:

Who owns this now?

That question exposes the difference between experimentation and execution. Many organizations have plenty of people who know how to launch an AI pilot; far fewer have leaders who know how to operationalize AI strategy across an enterprise.

AI Strategy: Day Two Is Where the Real Costs Begin

One mistake I see repeatedly is that organizations dramatically underestimate what happens after deployment. The presentation usually ends with the successful implementation, but the real work starts the following morning.

Models require monitoring. Data changes over time. Integrations need maintenance. Business processes evolve. New regulations emerge. Employees require continued training. Technical debt accumulates quietly until someone finally has to address it.

Those aren’t exciting topics, but realistically they’re where most of the cost lives.

I’ve learned that if a financial model doesn’t include day-two operations, it probably isn’t an accurate financial model. Technology is never a one-time purchase; it’s an ongoing operating commitment.

Governance Isn’t Slowing You Down

One of the biggest misconceptions I encounter is that governance limits innovation. In reality, strong governance is what allows organizations to scale confidently. Without governance, every AI initiative becomes a one-off experiment with unclear ownership, inconsistent standards, and unpredictable outcomes.

The failures I’ve witnessed over the years were rarely caused by the technology itself. They were leadership failures. No one clarified decision rights, no one established accountability; no one defined success before the project began.

Governance creates those guardrails. When leaders understand who makes decisions, who accepts risk, and how success will be measured, organizations move faster and more efficiently because uncertainty disappears.

Accountability Changes Everything

One lesson every CIO eventually learns is that not every technically correct decision is organizationally possible. Boards have priorities; CEOs have competing investments; audit committees have risk concerns; investors have expectations. Every technology decision exists inside those realities, and people who haven’t been accountable to a board often underestimate how those pressures influence decision-making.

Technology leaders don’t simply evaluate whether something is technically feasible — they evaluate whether it’s strategically defensible, which is not the same thing at all.

I’ve spent significant time supporting mergers, acquisitions, and technology diligence, and those experiences have reinforced an important lesson in my mind: When organizations are evaluated, AI initiatives receive much more scrutiny than marketing presentations. The AI initiatives with clear governance, measurable outcomes, and executive ownership survive diligence. The science projects rarely do.

It’s all about accountability: How is success measured? What results is this initiative delivering? Who owns the outcome?

Discipline Is the Competitive Advantage

AI is exciting, but comes with its share of unrealistic expectations. The companies that continue chasing demonstrations of AI’s potential, regardless of practical considerations, will end up struggling in the long term. The organizations that focus on governance, operational excellence, executive accountability, and measurable outcomes will separate themselves from everyone else. That sequence is the core of the 7 Pillar framework I use with clients: strategy clarity and leadership alignment come first, and model selection comes much later.

The next generation of AI leaders will emerge from the companies that treat AI as an operating discipline instead of a technology experiment. That’s also the group I believe will define the next decade.

In the end, successful AI comes down to building organizations that know how to own the outcome. These are the questions my AI Practice at The Doyle Group works through with executive teams every week. If they sound familiar, let’s start the conversation.