AI Adoption Culture: Why the First AI Use Case Should Be the CEO
Updated: Nov 11
Most AI programmes start on the shop floor and then wonder why so little changes. I start in the executive office. The first AI use case in any business should be the person running it.

FOR THE BOARD: THE ONE-MINUTE VERSION
More than half of UK firms now use AI, yet only about 6% of organisations in McKinsey's latest global survey say it's making a real dent in profits. That gap is about culture far more than software, and culture is set by what the boss is seen to do on a wet Tuesday, whatever the strategy document says.
My answer is to begin with the CEO's own information needs. Free up their thinking time, let people watch it happen, and attitudes shift faster than any training programme could push them. The harder work comes after that, and it's where most wasted AI spend ends up.
Over half of UK firms use AI, but few lead any differently.
Access stopped being the problem a while ago. When the British Chambers of Commerce surveyed businesses in January 2026, 54% of the 668 firms that answered were using AI, up from around a third a year earlier. Most employ fewer than 250 people.
Value is another matter. McKinsey's State of AI in 2026 classes about 6% of respondents as high performers, meaning AI accounts for at least 5% of their EBIT. Nearly nine in ten use AI somewhere in the business, yet only 37% can point to any positive effect on EBIT.
Plenty of activity, then, and not much to show the bank manager.
What I see working with Scottish boards
I've been founding and leading businesses for over 30 years and building AI products since 2012, including edge AI hardware patented in the UK, EU and US and scaled across nine countries. The lesson I learned the hard way is that the order you do things in matters more than the kit you buy.
Most of my week now goes on founders, managing directors and boards at the point where an AI strategy meets the reality of their business. That's included executive readiness workshops and one-to-one mentoring for SME and social enterprise leaders on publicly funded programmes across Scotland. I've advised businesses from a family legal practice to a construction firm turning over more than £30m. And I run my own AI Growth leadership sessions for senior teams.
The sectors change from room to room, but the pattern rarely does. Firms that make progress with AI are usually led by people who changed their own working habits before asking anyone else to.
Why the CEO should be the first AI use case
Before AI goes near the wider team, point it at the information needs of whoever runs the business. Of everything I argue, this gets the most pushback.
Think about how a chief executive's week really goes. Michael Porter and Nitin Nohria of Harvard Business School tracked 27 of them across more than 60,000 hours of work. They averaged 62.5 hours a week, and meetings swallowed 72% of it, around 37 a week. Strategic thinking gets the scraps.
Meanwhile, most of what a CEO needs to know arrives late, filtered and in someone else's format: management accounts, board packs, long email threads, the customer complaints nobody got round to reading.
That's where I'd aim AI. Have it pull the week's numbers together and flag what's moved, summarise the customer feedback, draft the awkward questions for the next board meeting. It's unglamorous work that hands hours back to the person whose thinking time is worth most. My playbook on using AI as a partner shows how a leader can treat it as a thinking companion rather than a glorified search engine.
The bigger return comes from being seen to do it. A CEO who uses AI openly, shows the team the prompt and the output and admits where it went wrong, sends a message no policy document can. Anyone who's served an apprenticeship will recognise it: the young ones watch how the gaffer handles a new tool long before they'll pick it up themselves.
Research points the same way. BCG's AI at Work 2025 survey covered more than 10,600 people, and where leadership support is strong, the share of employees who feel positive about generative AI climbs from 15% to 55%. Only about a quarter of frontline staff say they get that support. McKinsey's high performers, likewise, are twice as likely as their peers to say senior leaders demonstrate commitment to AI.
I don't know of a study that tests CEO-first sequencing head to head. It's my judgement, built on the research into leadership behaviour and on what I've watched happen in boardrooms, and I'll put my name to it. I made a related case in my response to the KPMG 2026 CEO Outlook: lead AI from the front, and listen hardest to the people furthest from it.
What it looked like in one UK plant
The closest thing I have to proof is recent mentoring work with a large-scale manufacturing plant, part of an international group. The site had a six-figure quality problem and a team that understood it well. Faults were recorded at the end of the line without their causes, off-quality was reported monthly in a pack built by hand, and AI wasn't used anywhere in production.
We started with the managing director's desk. I built a prototype dashboard from the plant's own production and defect records, and he confirmed it was the view the site needed, one that would remove a large amount of manual work and create value in the data hidden in silos. He made it the first priority, with a single agreed measure to re-check at 90 days.
The same plan has him write a one-page AI manifesto for every employee: why the site is using AI, what's allowed, what's not, and that it's there to support operators while people stay accountable for what they produce. It's early, and the 90-day numbers aren't in yet.
What I'd do next is give every employee an anonymised voice through Voxant voice agents www.voxant.ai so the leadership can see resistance before it surfaces and build new values around what people say.
Voice does a second job too. It lets staff release the hidden data, the problems and know-how that pass by word of mouth and never reach a system. From there, enablement and change get shaped around new value for customers, with AI starting as augmentation inside a strategy that treats data as the start of new models and products.
That sequence is what I call the 4 Vs: the values, voice, value creation and vision a leader sets. Culture grows from those long before any policy reaches it.
What gets in the way of AI adoption culture
Starting with the leader gets you moving. Three things usually slow you down after that, and none of them is technical.
Your business has more than one culture
Leaders talk about "our culture" as though there's only one. There are several.
Edgar Schein of MIT Sloan shaped how most of us think about organisational culture. In 1996 he argued that three subcultures matter most: operators who keep the work moving, engineers and technical specialists who design the systems, and executives whose world is finance and the board. When they drift apart, he said, organisations stop learning and new ideas don't last.
Any Scottish SME will recognise the picture. In finance, one mistake can follow you for years. Sales runs on relationships, so anything that sounds automated feels like a threat to the customer. The site or operations team holds know-how that lives in people's heads and nowhere else. Then there's the handful of enthusiasts who've used AI at home for a year and are miles ahead of their colleagues.
Announce "we're adopting AI" and each group hears something different. You can see it in the numbers. BCG found regular AI use among frontline employees has stalled at 51%, while more than three-quarters of leaders and managers use it several times a week. McKinsey found 47% of middle managers and individual contributors report AI-related strain, against 31% of executives.
One rollout plan for every department is like issuing the same boots for the office, the warehouse and the building site, then acting surprised when somebody gets hurt. Map the subcultures before the use cases, and ask each one plainly what AI threatens for them and what it might give them.
Your team knows things you don't
In 1955 two psychologists, Joseph Luft and Harrington Ingham, drew a simple window with four panes to explain self-awareness. They called it the Johari window, after their first names, and it explains a lot about why AI programmes go wrong.
One pane is open. It holds what you and your team both know, like the approved tools and the pilot everybody's heard about. Another is hidden: what you know and haven't said, such as where you expect AI to take cost out, or that you're not all that confident with it yourself. Leave that unsaid and the rumour mill fills it in for you. A third is unknown to everyone, like what AI will do to your market in five years.
The fourth is the one that catches leaders out. Your blind pane is what your team knows and you don't, and it's usually bigger than you'd expect.
McKinsey asked US executives how many employees used generative AI for at least 30% of their daily work. They guessed 4%. Asked directly, 13% of US employees said they did. A worldwide study of more than 48,000 people by KPMG and the University of Melbourne found 57% of employees hide their AI use and pass off AI-generated work as their own. Almost half admit using it in ways that break company policy.
It's the reason I argue discovery is the work, not a phase to rush. Most AI programmes come unstuck in rooms the leadership never visits.
The aim is to make the open pane as large as you can. Feedback shrinks your blind spot, and candour shrinks the hidden one. A CEO using AI in plain view, mistakes and all, is the most practical disclosure I know.
People decide they can't learn this
Carol Dweck's research at Stanford separates two mindsets. In a fixed mindset, ability is something you have or lack. In a growth mindset, it's built through effort and learning.
Writing in Harvard Business Review, she reported that employees in growth-mindset companies feel far more empowered and committed, and get more backing for collaboration and innovation. In largely fixed-mindset companies, the one thing people reported more of was cheating and deception. Her work with the consultancy Senn Delaney across seven Fortune 1000 companies found growth-mindset employees agreed 65% more strongly that their company supports risk-taking.
No technology I've worked with tests mindset as hard as AI. When a director tells me "I'm not really a tech person", that's a fixed mindset talking. You hear it too in the quiet fear of the senior professional who spent 20 years building expertise and now watches a graduate get more out of a tool than they can.
A leader who learns in public gives the whole business permission to be a beginner again. One who won't sends the opposite signal, however generous the training budget. As I wrote in Human-AI Collaboration: Why AI Often Adds Work Instead of Saving It, a culture that says it values judgement has to give people the time, authority and confidence to use it.
If you want to test yourself, Ryan Gottfredson has extended Dweck's work to four sets of mindsets, and his free Personal Mindset Assessment takes five to ten minutes. Do it before your next leadership meeting and ask your senior team to do the same. Then compare results around the table. The gaps get hard to ignore.
Five tripwires that burn cash and time on AI
These are where I see capable leadership teams lose money, and every one of them is a culture problem wearing a project's clothes.
1. Tool theatre. The licences are bought, the demo gets a round of applause and a usage chart turns up in the board pack. It looks like progress and is mostly performance. Before a purchase order is signed, I ask leaders to say in one sentence what value they're after.
2. The pilot with no exit. A pilot that can't fail can't teach you anything, which is why I've argued the pilot phase is officially over. Agree the route to scale and the stop conditions before it starts. With clients I use "not yet" rather than "stop", because leaders hear "stop" as failure and will keep funding a dud rather than admit it.
3. Decreed, not believed. Adoption mandated from above fades as soon as attention moves on. Adoption that grows out of a pain the team already feels sticks. If your people can't name the problem AI solves for them, expect them to comply in meetings and work round it everywhere else.
4. Silence mistaken for safety. No complaints doesn't mean no risk. More likely, nobody's telling you. I ask every leadership group I work with whether they could list everything their AI tools can reach. Try answering that before your next board.
5. Oversight assumed. An agent doesn't just answer, it acts, and that changes the leadership job. For each use case a named leader decides whether a person approves every action, keeps watch, or sets the goals and limits. Leave it undecided and you've handed your judgement to software by default.
How to build AI adoption culture before you spend
My client work rests on three frameworks of my own: the 4 Vs, the 4 Cs of AI Governance and CHANGE. I won't unpack them here, because they only come alive applied to a real business. All three start from the same place. Culture is the foundation, and governance, controls and skills go on top.
BCG puts numbers on that. By its reckoning only about 10% of the value in an AI transformation lies in the application and 20% in data and technology. The other 70% is down to workflow redesign, culture, governance and how people and AI work together. McKinsey's high performers fit the pattern: nearly three-quarters have fundamentally redesigned workflows because of AI, against roughly a quarter of other firms.
The first steps cost less than most leaders expect:
Start with your own information needs and use AI on them where people can see.
Run an anonymous check on which tools your team uses and what for.
Put shortlisted use cases through a challenge round. Invite colleagues to pull them apart, then sort what comes back into safeguards and stop conditions.
Write one page on what AI is for in your business and what it isn't for.
Protect learning time in the diary. BCG found regular use is far higher among people with at least five hours of training plus in-person coaching, yet only about a third of employees feel properly trained.
Sometimes the right advice is to hold back. On one care-technology engagement involving sensitive children's data, I recommended taking AI out of the first version of the product. Data flows, access controls, audit trails and a Data Protection Impact Assessment came first, with AI to follow once those foundations could carry it.
This is the work where 360 Strategy provides AI consulting in Scotland, helping boards and owner-managers get culture and guardrails right so the money that follows has somewhere useful to go.
Five questions to put to your leadership team
Has the CEO used AI on their own work, in front of the team, in the last month?
What does your team know about AI use here that you don't?
Which department would hear "we're adopting AI" as a threat, and has anyone asked them?
How much of this year's AI budget is going on people rather than licences?
Which of your current pilots has an agreed "not yet" condition?
If most answers are "not sure", look at your leadership habits before your technology.
AI Growth: the new growth mindset for leaders
Scotland has no shortage of AI tools, and they get cheaper every quarter. Leaders who've changed how they think and lead because of AI are far harder to find.
My AI Growth: The New Growth Mindset sessions are built for that gap. They're for founders, managing directors and boards who'd rather lead AI adoption than chase it. We work on making yourself the first use case, reading your subcultures and shrinking your blind spots, and you leave with decisions to take straight back to your business.
Ready to stop paying for tool theatre? Book a Clarity Call.
Closing thought
Tools can be bought in an afternoon. Culture takes far longer, and it's the one part of your AI strategy a competitor can't copy by signing the same contract.
Start with yourself. Your people are watching either way.
Mark Evans is founder of 360 Strategy, an AI readiness and governance consultancy based in Scotland.
Sources
British Chambers of Commerce and University of Essex: AI Adoption and Workforce Change in SMEs, ISER Working Paper 2026-01, March 2026
McKinsey: The State of AI in 2026: On the Road to ROI, August 2026
McKinsey: Leaders Underestimate Employees' AI Use, March 2025, drawing on Superagency in the Workplace
BCG: AI at Work 2025: Momentum Builds, but Gaps Remain, June 2025
KPMG and the University of Melbourne: Trust, Attitudes and Use of AI: A Global Study 2025
Michael Porter and Nitin Nohria: How CEOs Manage Time, Harvard Business Review, July-August 2018
Edgar Schein: Three Cultures of Management: The Key to Organizational Learning, Sloan Management Review, 1996
Joseph Luft and Harrington Ingham: The Johari Window, a Graphic Model of Interpersonal Awareness, UCLA, 1955
World Economic Forum: Growth vs Fixed Mindset: Which One Does Your Company Have?, November 2016, reporting Carol Dweck's research
Ryan Gottfredson: Personal Mindset Assessment
Comments