Most agencies didn't decide who would own AI adoption. It just landed on someone, usually whoever already owned "people stuff," because AI touches how people work.

New research from The Coders Guild puts a number on what that's doing to them. Ninety-five percent of UK HR leaders report rising workloads as AI adoption accelerates. Only 24% of organisations have a documented AI workforce strategy. And just 23% of HR departments have received AI-specific training themselves, despite being expected to lead the rollout for everyone else.

Read that last figure again. Fewer than a quarter of the people responsible for guiding their organisation through AI adoption have had any structured training in it. They're being asked to write the map while walking the route for the first time, at the same pace as everyone following them.

This isn't a story about HR being overwhelmed, though they are. It's a story about sequencing. Somewhere between "we should use AI more" and "here's our AI strategy," a step got skipped: giving the person leading adoption the grounding to actually lead it.

What this looks like in an agency

Agencies are a good example of where this bites hardest. Teams are small enough that AI ownership tends to fall on one person by default rather than by design, usually whoever runs operations, people, or both. That person is now fielding questions about which tools are safe to use on client work, how to brief AI without leaking confidential material, and what "good" looks like when a junior submits AI-assisted work. All reasonable questions. All hard to answer well without training.

This was close to the starting point at The Turner Agency, a 36-person global events and film agency. Before any training was designed, Go Wisely ran a full AI readiness diagnostic across six dimensions, including People and Culture and Governance, the two areas where this exact gap tends to hide. The diagnostic didn't assume the answer. It found where the actual gaps were, then built a six-session bespoke AI Accelerator training programme around them. Across 131 survey responses over the course of the programme, satisfaction landed at 4.6 out of 5, and not one response was dissatisfied.

The reason that worked isn't that the training was unusually good, although it was well received. It's that the sequence was right. Nobody was asked to lead adoption before they understood what they were leading people towards.

The fix is smaller than it sounds

The instinct, when a gap like this surfaces, is to assume it needs a large intervention: a hire, a consultant on retainer, a six-month programme. Often it doesn't. It needs someone to sit down for half a day, map where the business actually stands across vision, opportunities, people, tools, governance and value, and hand back a plan that says what to do first.

That's a different exercise from training. Training teaches people to use AI well once you know what they need to learn. A diagnostic tells you what that is before you spend a penny on delivering it. Skipping straight to training, which is the instinct most businesses follow, is how you end up with generic prompting workshops that don't touch the actual bottleneck: whether the person expected to lead adoption has been equipped to do it.

If AI ownership has quietly landed on someone in your business without anyone checking whether they're set up to carry it, that's worth naming honestly rather than working around. It's fixable, and it doesn't take long to find out what fixing it actually requires.

Go wisely.