The most common AI rollout still looks something like this: leadership buys a batch of licences, sends an email announcing "we're using AI now", and waits for productivity to follow. A few weeks later, adoption is patchy. Some people have quietly gone all in. Others have opened it twice and closed the tab. A few are actively worried about what it means for their job, and nobody's really talking about that part out loud.
None of this is a failure of the tool. It's what happens when access is treated as the same thing as readiness. Giving someone AI doesn't make them AI-ready any more than giving someone a gym membership makes them fit. Readiness is built, deliberately, over weeks — not switched on the moment a licence is issued.
Two different gaps, and they're not the same fix
There's a capability gap — people genuinely not knowing how to get good results from these tools, which is real and fixable with practical training. And there's a trust gap, which is different: uncertainty about whether it's safe to use, what the boundaries are, and whether being good at using AI is a threat to their role or an asset to it. Training alone closes the first gap. It rarely touches the second.
Start with the willing, not the whole company
A full-company rollout on day one usually surfaces every version of resistance at once, which is hard to manage well. It's often more effective to start with a smaller group who are genuinely curious, give them the training and the trust-building conversation together, and let their results do some of the persuading. Visible, credible internal examples tend to shift a wary team faster than another all-staff announcement ever will.
Why the trust gap matters more than it looks like it should
People who don't trust the "why" behind an AI rollout tend to do one of two things: avoid the tools altogether, or use them carelessly, without much thought for what should and shouldn't go into them. Both outcomes look the same from the outside — low, uneven adoption — but they come from opposite problems, and neither one gets fixed by a longer feature list.
What leading it well actually involves
This is closer to change management than IT deployment: leaders working through the change themselves first, building and protecting trust deliberately rather than assuming it, leading with empathy without losing clarity about what's expected, and being ready to handle resistance without treating it as a discipline problem. Teams that are led through this well tend to end up using AI more consistently and more sensibly than teams that were simply handed the tool and left to figure it out.
Capability and trust, built together
The two don't need to happen in strict sequence, but they do need to happen in the same programme rather than as separate initiatives run by different parts of the business. A capability session that never addresses the "is my job safe" question underneath it will produce technically competent staff who still hold back. A trust conversation with no practical follow-through leaves people reassured but no more capable than they were before. The combination is what actually shifts behaviour.
What resistance actually sounds like
It rarely announces itself as "I don't trust this." It sounds like a task quietly not getting delegated to an AI tool that could easily handle it, or a team member who says they're "too busy to look into it" three months running, or someone technically compliant with a new AI policy while privately doing everything the old way. None of that is defiance. It's usually an unaddressed worry — about job security, about looking incompetent while learning something new in front of colleagues, or about being blamed if an AI-assisted piece of work goes wrong. Naming that directly, rather than treating it as a training gap, is usually what actually moves it.
Where a workshop fits in
This is exactly the ground our Building trust in an AI world workshop is built for — for leaders and managers bringing their teams through AI-driven change, rather than announcing it and hoping. It sits alongside our more hands-on AI capability sessions, and which one a team needs first depends entirely on where they're starting from. Some need trust before tools. Some are already prompting confidently and just need the tools to go further.