What Britain's biggest companies are learning about AI adoption, and what small businesses can steal from it

Short answer: BT's own experience rolling out AI to its workforce points to one conclusion. Training gets people started, but it's hands-on use that actually builds the skill. Julian Tan, Group Director of Corporate Strategy at BT, puts it plainly: "The utility of actually using the tool is where a lot of the learning comes from." The scale is different for a 20-person business, but the mechanism isn't.

What Julian is learning, in his own words

Large employers have the kind of resources most small businesses will never have: dedicated learning programmes, internal platforms, a workforce large enough to run proper pilots. So it's worth paying attention when Julian says formal training isn't actually the thing doing most of the work.

"There is only so much that training can afford you in learning about the tool and what it's capable of doing," he says. "The utility of actually using the tool is where a lot of the learning comes from. So whilst at BT there are programs that allow our employees to learn more about the technology and how to use it, what we're seeing is the biggest driver of using these tools is actually getting the tools into the hands of our workforce and having them play around with it, experiment with it, and in some ways go through their own journey of exploring the art of the possible."

That's a striking admission from a company with the budget to run as much formal training as it wants. If hands-on use beats training even at that scale, it's not a shortcut small businesses are taking by skipping elaborate programmes. It's the actual mechanism that works.

The three things that actually drive adoption

Julian breaks down what drives adoption into three ingredients, none of which require large-company resources to copy:

1. Access. People need the tools in their hands, not just a demo or a one-off session.
2. Time and space to experiment. Adoption doesn't happen if using the tool is squeezed into five minutes at the end of a busy day.
3. Role modelling and myth-dispelling. Julian is direct about the fear this needs to counter: "there is a misconception that people worry AI is going to replace their jobs." Someone visible has to work against that, not just say it once.

He frames the underlying goal as reprogramming how people think about the tools altogether: "What we're doing is really trying to re-engineer the way that people think about how to use these tools and how these tools can help each of us do more than we are already capable of doing." And crucially, he doesn't treat any of this as a large-company problem: "I think that piece of how we're really driving AI adoption is relevant regardless of whether you're a large organization or a small one."

Translating this down to SME scale

A 20-person business can't run a corporate learning platform. But it can do the three things above, and arguably do them faster than a large company can, because there are fewer layers between "decide to try this" and "everyone's actually using it."

• Access means making sure your team has an actual paid AI account, not "ask Dave if he'll try it on his own ChatGPT login." (For more on why paid, company-provisioned access matters specifically for security reasons, see AI security on no budget.)
• Time and space means blocking out an hour, not hoping people find fifteen minutes between meetings. This doesn't need to be formal, it just needs to be protected.
• Role modelling means the owner or senior manager is visibly using the tool themselves, not just telling everyone else to.

None of that needs a training budget. It needs a decision to prioritise it.

What to try this month

Pick one recurring task your team already does, such as drafting client emails, summarising meeting notes, or first-pass proposal writing, and give everyone an hour this week to just try doing it with AI, with no expectation of a polished result. The point isn't the output. It's the repetition Julian describes: the more people actually use the tool on real work, the faster they get good at it.

For a free, structured way to build this into an actual plan rather than a one-off, the AI for Growth community gives small businesses access to the Reskilling Navigator and a peer network working through the same adoption curve.