Gallup's newest research on workplace culture makes an argument that's easy to agree with in principle and easy to skip in practice: AI itself doesn't build a better workplace. Managers do.

The numbers are stark. Ninety-nine percent of HR leaders say AI is important to their strategy. But half of them don't have confidence in their managers' ability to actually guide employees in using it. And the gap between those two positions shows up directly in how people work. Where employees feel their manager actively champions AI, a third say it's transforming how work gets done. Where they don't feel that, the figure drops to 4%. Employees with AI-supportive managers are also far more likely to report their culture improved this year, 31% versus 21%.

The training data underneath those numbers explains why. Only around half of organisations offer role-specific AI training, and just a quarter embed AI skills into job expectations or performance goals. Twelve percent have no formal enablement plan at all. So the managers being asked to champion AI use, in most organisations, haven't been given the tools to do it credibly.

Why this hits agencies particularly hard

Agencies run on a flatter structure than most businesses their size. A creative director, an account lead, or an ops manager is often the only layer between "the agency has decided to use AI" and "here's what that means for your actual client work." That person's confidence, or lack of it, becomes the whole organisation's AI culture by default.

If that manager is unsure which tools are appropriate for client-facing work, unclear on what "good" AI-assisted output looks like, or simply hasn't had the chance to build fluency themselves, the message that reaches the team is uncertainty, not direction. Gallup's data suggests that's not a minor drag. It's close to the entire difference between AI adoption that changes how work gets done and AI adoption that stalls at a few people quietly experimenting on their own.

The instinct, when this gap shows up, is to fix it with a policy document or a company-wide announcement. Neither addresses the actual mechanism Gallup identified, which is manager confidence built through real capability, not top-down instruction. A memo doesn't make someone a credible AI champion for their team. Understanding the tools well enough to answer real questions does.

What this looks like done well

This is precisely why a capability plan, built from an honest read of where an organisation's people actually stand, matters more than a generic training rollout. Blanket prompting workshops rarely reach the specific gap Gallup is describing, because they're not designed to identify which managers need what, or why. A structured diagnostic does that work first: it maps where confidence is genuinely low, where it's just untested, and what a realistic path to competence looks like for the people who will be expected to lead by example.

Gallup's framing is worth sitting with because it removes an excuse. The technology isn't the bottleneck. Management capacity is, and management capacity is buildable, deliberately and relatively quickly, once you know where it's thin.

If AI adoption in your agency depends on a handful of managers who've never had proper support to lead it, that's a findable, fixable gap, not a permanent weakness in the team.

Go wisely.

Source: Gallup, "The AI Effect on Workplace Culture," published 17 August 2026.