In most agencies, a few people are already deep into AI and everyone else is getting on with the job. PwC's latest workforce survey puts numbers on that split, and the numbers are worth a slow read.

The Global Workforce Hopes and Fears Survey 2026, published on 29 September, asked 49,364 workers across 48 countries and regions in May and June. PwC sorts people into groups. The "front-runners", 14% of the workforce, have scarce skills and strong AI capabilities. The largest group, 56%, is what PwC calls the "engine room": people with neither scarce skills nor substantial AI experience.

The gap between the two is not only about skill. Only two in five engine room workers say they can get the learning resources they need. Across the whole sample, 64% had used AI at work in the past year, up 10 points, and 22% now use generative AI daily, up from 14%. Yet the share who can access the training they need fell from 59% to 51%. Use is rising faster than support.

There is a retention point tucked in as well. Among the front-runners, 29% say they are very or extremely likely to change employer within 12 months. PwC's global workforce leader, Pete Brown, put the risk this way: "There is a real risk that the global workforce is starting to move at different speeds."

Three cautions before reading this into your own business. It is a global survey and no UK breakdown appears in the coverage, so it is not a measure of UK agencies. It measures workers' own reports, not results. And the segments are PwC's construction, so treat the percentages as a picture, not a benchmark.

Even so, the shape will look familiar to most agency owners. Someone on the team has been using AI for months and gets quietly better. The producers, project managers, account handlers and finance and operations people who keep clients served week to week are, by and large, waiting to be shown something that fits their day. When a training offer does arrive, it tends to be an all-hands session pitched at the middle, which suits the front-runners least and the engine room barely more.

The gap does not close by itself, and it is not a talent problem. It is a support problem, which makes it a fixable one.

What tends to help is small and specific. Start by finding out where people actually are, whether through a structured diagnostic or simply asking. Then choose one real task per role, something from this week's work, and practise on that. Keep sessions short and spread out, so people try things between them and come back with questions. And give your quick adopters a job that is not only their own output: showing a colleague what worked is better for both of them, and it gives them a reason to stay.

For an agency owner, the most useful question this week is a plain one: who on my team has been shown something useful for their own job, and who has been left to work it out? The answer usually shows where the next hour of effort is best spent. It does not call for a big programme or a new tool. It calls for attention to the people who quietly keep the place running.

Go wisely.