You can roll out AI across an organisation in a week. You buy the licences, switch on Copilot or ChatGPT, run a few demos and publish a sensible guide on how to use it.
Getting people to actually change how they work takes a lot longer. Most organisations we speak to have done the rollout. Far fewer have seen people genuinely work differently six months on.
That gap isn’t a technology problem. It’s a behaviour change problem. And we’ve been helping organisations with it long before anyone heard of a large language model.
The rollout isn't the hard part
When leaders talk about AI adoption, they usually mean access. Who has a licence. Which tools are approved. How many people logged in last month.
But people logging in doesn’t mean people working differently. A team can have every tool going and still draft every report from scratch, double-check everything by hand, and quietly avoid the new system when the pressure’s on.
The tools are the easy bit. The habits, beliefs and behaviours of your people are the hard bit. And those habits, beliefs and behaviours decide whether you see any return at all.
Change and transition: why people get stuck
It’s not change that trips us up. It’s the surprise, uncertainty and transition that come with it.
A useful way to think about it:
- Change is external. A new tool, a new process, a new way of being measured.
- Transition is internal. How people think about it, feel about it and respond to it.
Most AI programmes put nearly all their effort into the change and almost none into the transition. Yet when you listen to people, they’re stuck in the transition:
- “I don’t trust what it gives me.”
- “It’s quicker to just do it myself.”
- “I don’t want to look stupid asking.”
- “If I get good at this, am I training my replacement?”
None of those blockers get fixed by a prompt writing workshop. They’re mental, emotional and social. And most people have had almost no training in those skills.
What the research says
The evidence points the same way again and again.
- BCG’s 10, 20, 70 model. BCG says organisations get about 10% of AI’s value from algorithms, 20% from technology and data, and 70% from people and processes. Most budgets run the other way round.
- Adaptability is the skill employers say they need, and they rarely train for it. Barclays LifeSkills research found 60% of employers say adaptability has become more important over the last decade, 20% say employees lack it, and only 8% provide specific training for it.
- Adaptability beats technical knowledge. A Harvard Business Review survey identified adaptability as the most important skill for companies undergoing digital transformation, ahead of technical knowledge.
- Habits take time. Behaviour-change research suggests people need about 66 days of repeated practice to form a new habit. A launch event and a half day workshop don’t come close.
Put simply, most organisations invest in the 30% and hope people sort out the 70% on their own.
Why AI training on its own doesn't stick
Knowing how to use a tool and changing how you work in pressure moments are two different things.
People leave a good AI training session informed and even excited. Then Monday arrives. The deadline’s tight, the inbox is full, a client is chasing, and people go back to what they know. Not because they forgot what they learned, but because they never built the skills to do it differently when it counts.
We see the same pattern in culture change, leadership development and mental health. Awareness tells people why. Skills give people the how. Without the how, the why is just noise.
Train people for the road, not the road for people
Most organisations spend their energy trying to build a smoother road for their people. Better guides, cleaner processes, more comms, another town hall.
We think that’s backwards. Train your people for the road ahead.
AI is the latest stretch of road, and it won’t be the last. The organisations that keep adapting won’t be the ones with the best tools. They’ll be the ones whose people can handle uncertainty, learn fast and stay engaged while everything around them shifts.
The skills that make AI adoption stick
These are the skills we see make the difference between a rollout and real adoption:
- Managing the stress response. When something new lands, people’s threat response kicks in. People who can recognise and manage that response stay in control of their thinking.
- Challenging unhelpful thinking. “I’m not a tech person.” “I’ll get it wrong.” “This will replace me.” People who can spot these thinking traps can choose a more useful response.
- Staying curious rather than defensive. Resilient people treat a new tool as something to experiment with, not something to protect themselves from. Resilience isn’t a personality trait. People can learn it.
- Having honest conversations. People need to be able to say “I don’t get this” or “I’m worried about my role” without fear. Teams who can do that adapt far faster.
- Building the new habit. Practice, reinforcement and support over weeks, not a single session.
- Coaching, not telling. Managers who coach their people through change, rather than tell them to get on with it, keep their teams engaged and moving.
How to build behaviour change into your AI rollout
If you’re responsible for HR, L&D or a transformation programme, here’s where we’d start:
- Start with people’s worries, not the tool’s features. Find out what’s actually stopping people. Trust, time and habit come up again and again.
- Define what good looks like day to day. What does using AI well actually look like in a team meeting, a client email or a monthly report? Work it out with the people doing the work.
- Train the human skills alongside the technical ones. Prompting skills matter. They won’t stick if people can’t manage the uncertainty that comes with them.
- Equip your managers to coach. Your line managers are the people who make or break adoption. Give them the skills, not just the briefing.
- Plan for 66 days, not one launch. Build in refreshers, reminders and peer support so the new way of working becomes the normal way of working.
- Measure behaviour and wellbeing, not attendance. Track how people feel and work before and after, using validated measures, not just how many people logged in.
What the results look like
We’ve delivered skills based resilience programmes to more than 200 organisations, and we measure every one using validated clinical tools. On average, the people we work with see:
- burnout risk fall from 44% to 18%
- clinical level anxiety fall from 23% to 11%
- 112 hours of productive time back per person, per year
- a £12 return for every £1 invested
RBC saw a 35% fall in attrition. Kingston Council moved from the worst performing London Borough for Adult Social Care to second best, with a return of £11.08 for every £1 invested.
None of those clients came to us for AI. They came to us because their people were dealing with constant change. AI is simply the latest version of that challenge.
Ready to get your people through the transition?
Frequently asked questions
Why do AI adoption programmes fail? Most AI adoption programmes focus on access to tools and technical training, while people’s habits, beliefs and behaviours stay the same. BCG estimates around 70% of the value of AI comes from people and processes, so programmes that skip the human side rarely deliver the return they promised.
What is the difference between change and transition? Change is external: a new tool, process or structure. Transition is internal: how people think, feel and respond to that change. People get stuck in the transition, which is why it needs its own support and training.
What skills do employees need to adapt to AI? Alongside technical skills, people need to manage their stress response, challenge unhelpful thinking, stay curious, have honest conversations about their concerns and build new habits through practice. Managers also need coaching skills to support their teams through the change.
How long does it take for a new way of working to stick? Research on habit formation suggests around 66 days of repeated practice on average. That’s why one off launch events and half day workshops rarely change behaviour on their own.
