Giving someone access to AI doesn’t mean they know how, when, or whether they should use it. Yet as generative AI arrives in organisations faster than people can adapt, the temptation is to buy the tools, switch them on and tell everyone to start using them.

We learned this the hard way.

Our early attempts at AI adoption showed us that the biggest barriers weren’t technical. They were human. People worried about getting things wrong, exposing organisational information or trusting AI-generated outputs. Others simply couldn’t see where AI could genuinely improve their working lives.

So we changed our approach, starting with people’s needs rather than the technology. We focused on building confidence and creating an environment where people had the skills to experiment safely. Training moved beyond using a particular tool to cover what AI can do, where its limitations are, how to protect sensitive information and how to identify opportunities in people’s own work.

Building that confidence also meant finding the right balance between experimentation and control. Lock AI down too tightly and employees never develop the confidence or experience required to innovate. Open everything without education and governance, and organisations expose themselves to unnecessary risk.

In this talk, we share what we learned about creating the space between those two extremes: removing unnecessary barriers so people can experiment, while giving them the knowledge and guardrails to do it safely.

We’ll explore how to identify adoption blockers and knowledge gaps, tailor training to different clients and needs, using practical experimentation grounded in people’s everyday work to build confidence in responsible AI use.

Successful AI adoption isn’t about forcing everyone to use AI. It’s about giving people the confidence, skills and permission to discover where it can genuinely help them work better.