AI “Transformation,” Part 4: The CEO’s Decisions

Everything lands on the CEO’s desk eventually, sometimes all at once, all the time.

If you’re just joining: our CTO, chin-deep in last year’s “transformation” priorities, was handed a 20-minute board slot to answer “What are we doing about AI?” In Part 2 she wrote two complete decks in one night, The Roadmap and The Reset. In Part 3 she realized both decks fail in the same five places, and that most of what needs fixing exceeds her authority. At 6:55am she texted her CEO: “We need to talk.”

This final installment takes The CEO’s perspective.

The Text

It’s 7:10am and the CEO is at his desk, the only hour of the day when things are quiet. The text doesn’t surprise him. If anything, it’s overdue. He was copied on the CFO’s email weeks ago, the one with the competitor’s investor deck and its claimed 20% efficiency gain, and he had replied quickly: “This is what we should be covering at the board.” He has thought about that reply more than once since. It was true, but his own questions about AI were far greater than 20 minutes could cover.

He types back: “Let’s meet at noon today.”

The Triage Instinct

Everything lands on the CEO’s desk eventually, sometimes all at once, all the time. Early in his tenure he wished his team would resolve more of it themselves. He has since accepted that the pace of the business, and the pressure on every functional leader to deliver their own numbers, means the hard conflicts end up in his inbox.

So he learned the discipline that he considers the most valuable behavior of his career: triage. If an issue is real and can be settled with a clear decision that needs no broader input, decide it now and move on.

That habit saves an enormous amount of time and angst within the company, even if the CEO never sees it. But if an issue raises deeper questions about the future of the business, a fast answer does as much damage as a slow. Most of the job is knowing which bucket you are in before you decide.

What the CTO Asks For

At noon she walks in with no laptop and no deck. He notices and is relieved; he has spent eight years trying to teach people that the slide is not the strategy.

She walks him through the five issues she wrote down earlier that morning: governance built for a slower world, the talent gap, strategic clarity at the board, whether the business strategy itself still holds, and the shaky foundations of process and data underneath all of it. He recognizes the list, and he has been contemplating two of the items for months.

Then she does something most of his executives never do. She doesn’t ask for guidance, she asks for decisions, five of them:

1. Poll the board before the meeting to surface the question behind the question

2. Commit that enterprise upskilling is an HR and business-led effort with IT supporting

3. An ELT offsite to stress-test the strategy against what AI makes possible

4. Create a second, faster governance lane with bridge funding that doesn’t wait for the annual cycle

5. Promote her top lieutenant to run everything already in flight (the legacy investments), so she can stop defending the past and focus on the future.

While she talks, he sorts the asks into his two categories.

Quick Decisions

Three of the five land in the quick decision-making bucket.

· The lieutenant: yes, on the spot. It’s the right move for the person, it’s probably a year overdue, and it costs little while buying the company the CTO’s full attention. “Done. Announce it this week.”

· Upskilling: yes, and he agrees with her framing. HR and the business own it, IT supports it. He adds one condition of his own: a baseline AI competency expectation for every employee, with usage tracked as a carrot, not a stick. Basic fluency with these tools will soon be as unremarkable as using the internet, and he wants no one pretending otherwise. But this question relates to a bigger issue he has been grappling with.

· The offsite: yes. It was already on the calendar, but he will change the agenda. That’s the easy part but he has deeper questions about strategy as well.

Ten minutes, three decisions. This is the part of the job he still loves! But now he must address the deeper questions.

The second governance lane? No.

He agrees with her diagnosis, which is what makes the answer harder. The planning process was built to prevent bad surprises from technology that evolved on a five-year cycle, and it is now being asked to govern technology that changes on a three-month cycle.

He also knows the reality: run the AI question through the current process and the board’s questions will be twelve months old before the first project starts, and twelve months is an eternity.

But he has seen what happens to companies that build exceptions to their own discipline. The fast lane becomes the lane everyone wants to drive in, and two years later he might have to explain big write-offs.

So he offers her something different. Bridge funding now, from budget he controls outside the departmental plans, reviewed monthly with him directly. And a recalibration of risk, audit, and compliance rather than an exemption from them: their mandate must adapt to acknowledge risks they don’t yet fully understand. A control function that restricts whatever it doesn’t fully understand isn’t reducing risk – it is just shifting it into the company’s competitiveness, where no auditor will ever notice it.

“AI is not an ‘initiative’, and I won’t govern it like one. It will be a consideration in every business case we approve from here on. Not a separate lane – a core expectation.”

She pushes back: without some structural protection, the current process will kill anything that moves at a different speed. She has seen it happen. He doesn’t dismiss it. The monthly steerco will be the test. If the current process fails, he tells her, he will build the second lane himself and steer it personally.

The issue isn’t fully resolved, but it’s resolved enough to get going.

Managing the Board – a Bigger Question than Planning the Agenda

Her first ask, polling the board, ends up in a different place: the 20-minute slot is cancelled entirely.

In its place, two hours on the first day, when the board handles committees. He will lead it, and she can offer her perspective on the risks and the opportunities. There will be no slides, because for the session to work people need to be truly engaged in the discussion not just reacting to slides.

What he doesn’t tell her is that the session is also a diagnostic. He wants to watch each director engage with the subject at length. He needs to know who in that room is over-reacting to risk and who can actually embrace opportunity, because he has a growing suspicion that the board that governed the last decade is not well suited for the next one, and fixing it means recruiting genuine technology depth, perhaps from outside the industry.

Evolving the Business Strategy

He doesn’t think her fourth issue, whether the strategy still holds, is an AI issue at all. That question should be asked and answered every quarter by his business leadership, and mostly it gets answered by reflex. Not this time: he will make it the core question of the next QBR, and the P&L leaders will answer it themselves, because the fastest way to doom every AI effort in the company is to let the business file it under IT. He tells her to expect knocks on her door as they prepare. “Put your consultant hat on. Help them see what’s coming. They can connect the dots from there.”

The Hardest Question

He saved the biggest issue for last, because he knows it is the one that may be the most disruptive.

“You called it the talent gap and asked who should own it. I think that’s a small version of bigger issue.”

His largest frustration across eight years as CEO has been the return on technology investment. The company wrote very large checks and realized questionable value. This has gone on so long that he has stopped blaming the IT organization for it (including hers.)

The problem, as he has come to see it, is an understanding gap between business leadership and technology leadership, and the chasm has widened every year as technology accelerated. Her fifth issue, shaky foundations, is the proof point. The scattered data, the spreadsheet reconciliations, the processes designed around human limits: those were never purely technology failures. They are what happens when the business treats its own data and its own processes as somebody else’s department. You cannot repair foundations from a central function, no matter how talented the CTO is.

So yes, she should upskill and reshape her team. That solves her problem. It does not solve his. Technology is becoming the fabric of the business itself, and building software is no longer the exclusive craft of specialists. Which leads him somewhere he knows she is not expecting to go.

“I keep arriving at the same place. I’m not sure this company should have a technology organization at all.”

He watches her predictable reaction and keeps going.

Technology embedded deeply in each business, owned by people who understand the business and own the results. Keep a small central core for what is genuinely common across business segments. And he adds: this will be harder on the business than on her. His P&L leaders have spent a decade building benches without weighing technology competency. That must change, fast, and they are going to have difficulty with it. Besides, he always felt she would make a great divisional COO.

Processing What Just Happened

The CTO leaves the meeting with more than she asked for. The promotion, approved in a sentence. The funding bridge. The offsite. A CEO who, it turns out, had been considering half her list already.

And yet the last ten minutes are a blur she cannot stop replaying in her head. Logically, she followed every step of it. Some part of her, the part that wrote “The Reset” at midnight, may even agree with it. But one question keeps troubling her:

What happens to the Chief Technology Officer when the CEO decides the company shouldn’t have a technology organization at all?

—Darryl