Most agencies now have an AI pocket. Somewhere in the business, one team has found a set of tools that genuinely works, and would not want to go back to doing the job without them. Almost always it is the content or creative team, sometimes project delivery. What is unusual is for that pocket to have spread anywhere else.

New research from Gartner, published on 1 September 2026, puts a number on how common this actually is. Surveying 1,303 organisations with revenues above $50m, Gartner found that only 22% have successfully scaled AI across multiple business units or taken what it calls an "AI-first" approach. The rest are somewhere earlier: piloting, experimenting, or contained within a single team. At the same time, 85% of functional leaders plan to increase their AI spending in 2026, having already committed an average of 12% of their functional budgets to it in 2025. Eleven per cent of organisations do not know what they spent on AI last year at all.

Read together, those numbers describe a specific and fixable problem. It is not that AI fails to work outside the team where it started. In most cases, the same drafting, summarising, and research tools that save a content team real time each week would do a similar job in project management, client services, or finance. What is missing is not capability, it is intent: nobody has been asked, deliberately, whether the tools that work here would also work there, and if so, what it would take to introduce them properly.

For agencies specifically, this pattern is easy to recognise once it is named. Creative teams tend to be the earliest and most enthusiastic adopters of AI tools, because the work lends itself to fast iteration and the payoff is immediate and visible. Project management and client services rarely get the same attention, not because the work resists AI, but because nobody has run the exercise of asking where else it might help. The result is a business that looks, from the outside, like it has embraced AI, while in practice one department is well ahead and the rest have not been invited to the conversation.

Part of the reason this happens is that nobody in a growing agency is explicitly responsible for the question. The team that adopted AI first did so on its own initiative, usually without a mandate to look beyond its own desks. Leadership, meanwhile, sees the enthusiasm in one part of the business and reasonably assumes AI is "being handled." Neither party is at fault, but the gap between them is exactly where the next 12% of budget risks being spent on more of the same, rather than on finding the departments that have not been asked yet.

This is precisely the gap the AI Accelerator Diagnostic is built to surface. At The Turner Agency, a 36-person events and film agency, the diagnostic found AI already well established within the creative team, but no clear view of whether project management or client servicing could benefit from the same tools, or what governance would need to look like if they did. The six-session training programme that followed brought all three areas into the same conversation, rather than leaving AI adoption to spread, or not, by accident. Across the programme, satisfaction averaged 4.6 out of 5 from 131 survey responses, with not one dissatisfied response recorded.

The lesson from the Gartner data is not that agencies should slow down on AI. It is that the next pound spent on it is better directed at understanding where AI already sits across the business, and where it has simply never been asked to go, than at deepening the pocket that already exists. Eighty-five per cent of leaders are about to increase their AI budgets again this year. The ones who answer the scaling question first will get considerably more from that spend than the ones who do not, and they will not need to guess at the answer, because it is exactly what a half-day diagnostic is designed to find.

Go wisely.

Gartner press release, "Gartner Survey Finds Only 22% of Organizations Have Successfully Scaled AI Across Multiple Business Units," 1 September 2026; survey fielded January-April 2026, 1,303 respondents from organisations with $50m+ enterprise-wide revenue