The CEO has to use AI personally, or the company won't

Microsoft published its 2026 Work Trend Index last week, a survey of 20,000 knowledge workers who use AI, across ten countries (Microsoft WorkLab). The finding I’d put in front of every CEO is this one: organizational factors like culture, manager support and talent practices accounted for 67% of the measured AI impact, against 32% for individual factors like a person’s own mindset and habits.

In plain terms, whether your people get value from AI depends far more on how the company runs than on how clever they are. And how the company runs starts with what the person at the top does every day.

The data on leaders who model it

The report draws on a separate Microsoft study of 1,800 workers. When managers modeled AI use themselves, employees reported a 17-point lift in the value they got from AI, a 22-point lift in how critically they thought about their own AI use, and a 30-point lift in trust in agentic AI (Microsoft WorkLab).

The report also found that only 26% of AI users say their leadership is clearly and consistently aligned on AI. Three out of four people are getting mixed signals from the top.

DataLift reported when managers model AI use

Value from AI17 pts
Critical thinking about AI22 pts
Trust in agentic AI30 pts

Employees whose managers used AI openly reported more value and more trust. Source: microsoft.com

These are survey results, and Microsoft sells the tools, so read them with that in mind. The report itself says these are associations, not proof of cause. They still match what you’d expect from twenty years of watching how companies adopt anything new. People copy what the boss does.

Delegating AI to the CTO is the common mistake

A common pattern goes like this. The CEO decides AI matters. The CEO hires or appoints someone to “own AI.” That person buys licenses, runs a lunch-and-learn and sends a usage dashboard every month. The CEO glances at the dashboard and goes back to writing board decks the old way.

The team notices. If the CEO doesn’t use it, AI is optional. If it’s optional, most people will keep doing their work the way that already works for them, because that’s the safe choice when you have a deadline.

BNY’s CEO Robin Vince understood this. In a memo to several thousand senior leaders, he urged them to model AI use, and he described the shift as “a fundamental leadership shift, not simply a capability shift,” according to Business Insider. BNY has been around for about 240 years. If its CEO is asking executives to use the tools in front of their teams, a 100-person company has no excuse.

The Work Trend Index has a number on this as well. Among the workers Microsoft classed as its most advanced AI users, 85% said their managers openly use AI, against 64% for everyone else (Microsoft WorkLab). The word that matters there is “openly.” Quiet use by executives doesn’t change anyone else’s behavior.

What “using it personally” looks like

Using it yourself doesn’t mean asking a chatbot a trivia question once a week. It means running part of your actual job through it, often enough that you know where it’s good and where it fails.

Here’s what that can look like in a normal week for a CEO:

Draft your board update or investor letter with AI from your own notes, then edit it. Note what you had to fix.
Before each one-on-one, have it pull the open items from your last three meetings with that person.
Give it a messy spreadsheet from finance and ask it to explain the three biggest changes from last quarter. Check its math.
Hand it a decision you’re stuck on and ask it to argue the other side.
Set up one small agent on your own inbox that sorts and drafts replies to routine requests, with you approving every send.

Expect the first two weeks to feel slower than doing the work yourself. You’re learning how to ask, what to check and where the tool is weak. By the third week, the tasks that suit it will start coming back faster than you could write them, and you’ll know which ones to stop trying.

The point is to build judgment. You can’t set a quality standard for AI-assisted work if you’ve never produced any. The Microsoft data shows this too: among the most advanced AI users, 83% said their managers set quality standards for AI work, against 57% for everyone else (Microsoft WorkLab). A manager can only set that standard after doing the work. The same goes for the CEO, whose standard everyone else ends up copying.

There’s a business reason too. Every vendor pitching you an agent this year will show you a demo that works. If you’ve spent a month using these tools on your own work, you’ll know which claims are real and which questions to ask. If you haven’t, you’re buying on faith.

Show the work, including the misses

Using it in private isn’t enough. Your team has to see it.

The most useful thing a CEO can do in a leadership meeting is put a real example on the screen: “Here’s the investor update I drafted with AI. Here’s what it got wrong. Here’s the time it saved.” The misses matter as much as the wins. They give your team permission to try things and report back honestly, instead of pretending the tools work perfectly or quietly not using them.

The Work Trend Index found 81% of leaders feel safe suggesting new ways of working with AI, against 67% of employees (Microsoft WorkLab). Leaders feel free to experiment. Their people don’t. The fastest way to close that gap is to experiment where your people can see it.

One thing to avoid while you do this: don’t turn adoption into a login count. A dashboard that shows who opened the tool this week tells you nothing about whether the work got better, and people learn to game it fast. Ask for examples of finished work instead. What did you do with it, how long did it take, and what did you have to fix? Those three questions tell you more than any usage report.

People copy what the boss does. If you don't use AI on your own work, your team won't either.

What to do this week

Pick three tasks from your own calendar this week: one writing task, one analysis task and one meeting-prep task.
Do each one with whatever AI tool your company already pays for. Time it and note what you had to correct.
Bring one example, with its mistakes, to your next leadership meeting and show it on the screen.
Ask each person on your leadership team to bring one example of their own the following week.
Make it a standing ten-minute item on the agenda for the next quarter, and keep going first yourself.