AI Adoption in Your Team - Why Licences Sit Unused and How to Get People Using AI

In a 2025 Gartner survey of 187 IT leaders, only 5% of organizations that had finished a Microsoft 365 Copilot pilot planned a larger deployment that year. Meanwhile, according to Gallup, the share of US employees using AI at work at least a few times a year rose from 21% to 40% in two years, but only 8% use it daily. Buying licences is the easy part of an AI rollout - the hard part is getting people to use them in their everyday work.

According to Gartner, only 5% of companies that finished a Copilot pilot planned a larger rollout, and just 8% of employees use AI daily. Five kinds of resistance and what works, the role of managers, AI champions, measuring adoption at three levels and a 90-day plan.

EY's Work Reimagined 2025 report, based on responses from 15,000 employees in 29 countries, shows the scale of the problem: 88% of respondents already use AI, but mostly for search and summarizing documents, and only 5% use it in ways that genuinely change their work. According to EY, companies lose up to 40% of the possible productivity gains as a result. In this post I cover why licences sit unused, which kinds of resistance hold a team back and what works against them, how to build an AI champions program, what to measure and how to plan the first 90 days.

Why licences sit unused

The most common scenario goes like this: the company buys an AI package, sends an email announcing access, runs one training session and waits. Three months later a handful of people who would have learned anyway use the tool regularly, and everyone else goes back to old habits. Research points to three main causes.

• No plan. According to Gallup, 44% of employees say their organization has started integrating AI into its processes, but only 22% see a clear plan for how that should work.
• Too little training. In BCG's AI at Work 2025 study of more than 10,600 people in 11 countries, only 36% of employees feel adequately trained. In EY's survey, just 12% say their AI training is sufficient.
• No support from managers. According to BCG, only 25% of frontline employees get enough guidance on AI from their leaders.

The problem rarely lies in the tool itself - usually nobody has shown people what exactly to use it for in their own work. The choice of package matters, but less than it seems. I compare Copilot, Gemini and ChatGPT Business in the post on AI packages for business.

A map of resistance and what works

Resistance to AI isn't one thing. Behind "people don't want to" there are several different concerns, and each needs a different response.

Two of these are often overlooked. According to Slack's Workforce Index from autumn 2024, 48% of desk workers would be uncomfortable admitting to their manager that they used AI for common tasks. The top reasons are feeling it's cheating and fear of being seen as less competent or lazy. If people hide their AI use, the company doesn't know what works and can't spread it. It's the same dynamic that leads to shadow AI, the use of personal tools outside company control - according to Microsoft's 2024 Work Trend Index, 78% of AI users were bringing their own AI tools to work.

Fear for one's job is real too, and it doesn't fade with time. In BCG's study, 41% of respondents worry their job could disappear, and among regular AI users the share is even higher. Saying "nobody will lose their job" without specifics doesn't work. What works is an honest message about which tasks will change and what the company plans to do with the time saved.

The role of managers

In BCG's study, the share of employees who feel positive about generative AI rose from 15% to 55% when they felt strong support from their leaders. Gallup sees a similar pattern: engagement among employees whose manager actively supports AI use is 48%, versus 30% for the rest. The most effective AI training is a manager who shows in a meeting how they prepared something with AI.

In practice that means three things: managers learn first, they have a specific AI task for their team in the calendar and they talk about AI in regular meetings, not just in training sessions.

An AI champions program

People learn most readily from colleagues who do similar work. That's the basis of an AI champions program: a few people in each department get early access, more training and time to help others.

Citi is a good example. Since 2024 the bank has built a network of about 4,000 volunteers called AI Accelerators, supported by a few dozen champions. According to reports, more than 70% of its 182,000 employees use the bank's approved AI tools, and using them isn't mandatory. In a smaller company the principle is the same, just at a different scale:

• One person per 10-20 employees, from different departments, not just IT.
• Voluntary sign-up, but with the manager's backing and time in the calendar, for example two hours a week.
• Early access and more training, including security rules and how to verify outputs.
• A specific task: find three repetitive activities in their department that can be sped up, and show them to the team.
• Regular champion meetups, where they share what works and what doesn't.

How to measure adoption

The number of active licences only tells you whether someone opened the tool. To know whether AI is changing work, measure at three levels.

The most common mistake is stopping at the first level. A company where everyone opens the tool once a week to polish an email has high activity but almost no impact. Treat self-reported time savings with care: according to BCG, 47% of regular users say AI saves them at least an hour a day, but it's worth confirming such estimates by measuring a few specific tasks. I show how to calculate automation returns in the post on AI automation ROI.

Rituals that build the habit

Training provides knowledge, but a habit forms only when AI becomes part of the team's normal rhythm of work. Simple rituals work well:

• Short team demos. Every two weeks, 15 minutes in a meeting: one person shows one task they did with AI, including what didn't work.
• A use case and prompt library. A shared place with examples from your own work, not from the internet: the task, the prompt, a sample output and notes on what needs checking. I cover writing good prompts in the post on prompt engineering for business.
• A questions channel. One place in your messaging tool to ask AI champions and share results.
• Challenge of the month. One type of task the whole department tries to speed up with AI, such as preparing a report or answering common enquiries.

Training and the Article 4 AI Act obligation

Since 2 February 2025, Article 4 of the AI Act has required companies that use AI systems to look after their staff's AI literacy. After the AI Omnibus (in force since 27 July 2026), the provision requires measures that support AI literacy, but mandates no specific level or certificate. I cover the details in the post on AI literacy training. From an adoption point of view it's good news: training has to be organized anyway, so it should teach people to use AI in their actual work, not just the theory.

BCG's data shows that time and format matter. Among people who received more than five hours of training, 79% use AI regularly, compared with 67% of those who received less. In-person sessions with room for questions work best. One generic company-wide session usually isn't enough - shorter sessions for each department, on their own tasks, work better. It's worth combining the safe-use rules with your company AI policy, so people know what's allowed before they start experimenting.

A plan for the first 90 days

Companies in some markets have a lot of ground to make up. According to Eurostat, in 2025 AI technologies were used by 20% of EU enterprises with at least 10 employees, ranging from 42% in Denmark to 8.4% in Poland and 5.2% in Romania. If you're still planning your first rollout, start with the post on where to start with AI implementation.

A step-by-step plan

1. 1.Pick 3-5 processes where AI should help, instead of rolling out a tool "for everything".
2. 2.Write a short AI policy with examples of what is and isn't allowed.
3. 3.Train managers first and give them a task to show their team.
4. 4.Build a network of AI champions: one person per 10-20 employees, with time in the calendar.
5. 5.Train departments on their own tasks, for at least a few hours, with room for questions.
6. 6.Launch a use case library and short team demos.
7. 7.Measure at three levels: activity, use cases and impact.
8. 8.After 90 days, decide what to scale and what to drop.

---

I help companies make sure the AI tools they've bought actually get used - from choosing processes and an AI policy, through department training and an AI champions program, to measuring results. I do this as part of AI training and AI consulting. Get in touch - I'll start by talking to a few people on your team to see what's holding back AI use today.

Worth reading next:

• AI Literacy - Mandatory AI Training Under Article 4 of the AI Act: Who, What, and How
• Copilot, Gemini, or ChatGPT Business - Which AI Package Should Your Company Choose (2026 Comparison)
• Shadow AI - Your Employees Are Using AI Without Your Knowledge and That's Your Problem
• Prompt Engineering for Business - How to Talk to AI So It Actually Delivers