September 29, 2026
How to Make AI Adoption Stick: The Human Side of Change
Catherine Toms
AI Expert | Educator | Consultant
AI adoption is primarily a behaviour-change challenge, not a technology rollout. Giving people access to AI tools and training them to use them doesn’t always mean adoption. The bigger challenge is creating the conditions for people to change how they actually work.
We’re often brought into businesses after the AI tools have already been rolled out.
The licences are in place. People have usually had some basic training. A few are using AI brilliantly. But across the wider organisation, adoption is still patchy.
And when we dig into why, technology is rarely the biggest barrier. It’s all the human stuff.
Why AI Adoption Feels Different:
AI understandably feels very different from a typical tech rollout.
A new system might change where you store information or how you complete a process. AI can threaten something far more personal: how people think about their expertise, their judgement, the quality of their work and, ultimately, where their value comes from.
That’s why AI adoption can’t be tackled through skills alone.
People need to know how to use the tools, of course. But they also need the right conditions to change established behaviours and build new ones into their everyday work.
We focus on three conditions that help make that happen: Safety, Space and Signals.
1. Safety: Make it safe to try
People won’t adopt AI if they’re unsure what’s allowed, worried about getting something wrong or concerned about what it could mean for their role.
So before skills training even starts, we build safety around the change.
That means:
- Clear guardrails. People need to know what they can and can’t do with AI in the specific context of their role.
- Permission to test and fail. Not everything people try with AI is going to work. That’s part of the process and the learning is valuable too.
- Leaders modelling the behaviour. When leaders are in the room, learning openly alongside their teams, we consistently see greater trust, engagement and active use.
But safety isn’t only about policies and permission. It’s also about the beliefs people bring with them.
We all have mental models about what good work looks like, where our value comes from and how work is supposed to be done. With AI, those might sound like:
- “Using AI is cheating.”
- “If AI can do this, what does that mean for my role?”
- “Good work should take time.”
- “My expertise comes from knowing how to do this myself.”
These beliefs aren’t always visible, but they can have a big impact on whether someone experiments with AI or defaults back to familiar ways of working.
It’s important you get the elephants in the room out in the open early. And people can openly share worries about quality, trust and job impact. We don’t guess what’s holding people back. We ask.
Before and during adoption programmes, we use surveys, discussions and activities to understand people’s hopes, concerns and motivations, and to gather feedback as adoption progresses.
The aim isn’t to pretend there’s no uncertainty. It’s to create enough safety for people to learn and change within it.
2. Space: Give People Time To Apply
AI is often the highest priority that’s no-one’s job. People are told they need to “work differently”, then sent straight back to the same pressures and deadlines. No wonder people are feeling burnt out and overwhelmed.
Here’s what we build in:
- Carved-out time, agreed with leaders upfront. Protected time is agreed before the programme starts, both for live sessions and for putting the learning into practice back at work.
- Opportunities to test, learn and validate on real work. People use their own tasks, try different approaches and build confidence through guided practice.
- Shrink the behaviour. Instead of telling people to “use more AI”, give them one small, specific action to try in their work. Something realistic and repeatable that helps them see the value quickly.
- Plan for BAU. The risk point is when the programme ends and normal pressures return. So we plan for that from the start: what people will keep doing, when they’ll do it and who will support them.
None of this is complicated. But if you don’t create space for already stretched teams to change how they work it’ll always fall down the to-do list.
3. Signals: Reinforce the Behaviour You Want
You can tell people AI is a priority. But they’ll pay much more attention to what the organisation actually measures, recognises and rewards.
Microsoft calls this the Transformation Paradox: employees are being asked to reinvent how they work, while the systems around them still reinforce the old way.
In its 2026 Work Trend Index, Microsoft found that 45% of AI users said it felt safer to focus on their current goals than redesign work with AI. Yet only 13% said they were rewarded for reinventing how work gets done with AI, even when the results weren’t met.
That’s a pretty big disconnect.
The good news is that organisations can change the signals they send. There are three things we look at with teams:
- What gets measured? If AI is meant to change how work gets done, measure the outcomes of that change: better workflows, time saved, fewer handoffs, improved quality and less rework.
- What gets recognised? Make the good stuff visible. Run showcases and team demos. Share before-and-afters, useful agents and redesigned workflows. Celebrate the people doing the work.
- What gets rewarded? Competitions, small incentives, internal awards, progression — whatever fits your culture. Give people some upside for experimenting with and improving the way work gets done.
If you want people to work differently, the systems around them need to reinforce that change too.
From Skills to Sustained Behaviour
AI adoption doesn’t succeed because of tech – it succeeds because of people.
It’s about consistent efforts to change established ways of working and building new ones that stick. And that takes more than rolling out licences and skills.
If you’re working through the same challenge in your organisation, speak to us about how we design human-first AI capability and adoption programmes that stick.
FAQs
Why is AI adoption difficult?
AI adoption isn’t just about learning new tech. It can challenge established ways of working, people’s confidence in their expertise and judgement, and how they think about the value they create. Building the skills is a great place to start, but people also need the right conditions to put them into practice.
What makes AI training effective?
The best AI training is practical, relevant and connected to people’s real work. People need opportunities to get hands on, apply AI to their own tasks, see what works first hand and build confidence through practice. It also helps when they have internal support and time to apply what they’ve learned back at work.
How can organisations improve AI adoption?
Build capability and adoption habits together. Give people the skills to use AI well, make it safe to experiment, create space to apply those skills to real work, and reinforce new behaviours through what the organisation measures, recognises and rewards.
How should organisations measure AI adoption?
Look beyond training completion or frequency of use. The more useful question is whether work is actually changing: are people improving workflows, saving time, reducing handoffs or rework, improving quality, or finding better ways to get things done?
Explore the other guides in our AI at Work series, covering the people, strategy and operating changes needed to scale AI across organisations.
Explore more AI resources
Practical guides, frameworks and tools to help you build AI capability, improve how work gets done and scale adoption across your organisation.
