I spoke with a team leader at a financial services firm last month who was visibly exhausted. Her calendar was packed with AI training sessions—not attending them, but supporting her team through them.

“I’m supposed to be their first line of support,” she said. “But I barely understand this stuff myself.”

This is happening everywhere, and L&D teams aren’t talking about it enough.

The Invisible Load

When organisations launch AI upskilling programmes, the official plan usually involves: corporate training sessions, vendor demonstrations, maybe some e-learning modules. Box-ticking exercises that satisfy compliance requirements.

The actual learning happens afterward, in the daily questions. “How do I use this for my expense reports?” “The AI gave me weird output—what did I do wrong?” “Is it okay to put client names into this tool?”

Those questions land on middle managers. Often managers who have no more AI expertise than their direct reports but feel obligated to provide answers anyway.

The Capability Gap

Here’s the uncomfortable reality: most middle management training focuses on people skills, project management, and communication. Technical upskilling rarely targets this layer specifically.

Senior leaders get executive briefings about AI strategy. Individual contributors get hands-on training for specific tools. Middle managers get caught in between—expected to translate strategy into practice while simultaneously learning the tools themselves.

One manager described it as “being expected to build the plane while flying it while also explaining the engineering principles to passengers.”

What Organisations Are Missing

The most successful AI rollouts I’ve seen explicitly support middle managers, not just individual contributors.

Manager-specific training sessions. Not the same content as their teams, but training focused on how to support learning, troubleshoot common issues, and know when to escalate.

Clear escalation paths. When a manager doesn’t know the answer, they need to know who does. A named contact, a Slack channel, something concrete.

Protected time for their own learning. Managers can’t support their teams’ development if they’re never given space for their own. This needs to be explicit and scheduled, not “find time when you can.”

Realistic expectations from senior leadership. If managers are expected to be AI champions, that needs to be acknowledged as additional work, not absorbed into existing role expectations.

The Risk of Ignoring This

Burned-out middle managers become bottlenecks for AI adoption. They stop answering questions enthusiastically. They start deflecting to IT or the training team for every minor issue. Their teams sense the frustration and become less likely to experiment.

The Jobs and Skills Australia projections about AI adoption assume a certain level of workplace capability building. That building happens (or doesn’t) at the middle management layer.

When organisations report that “employees aren’t adopting AI tools as expected,” it’s often because the support infrastructure was never designed to handle the real learning curve.

The L&D Opportunity

Learning and development teams have a chance to position themselves as genuine partners to middle management, not just content providers.

That means asking managers what support they actually need, not assuming. Running debrief sessions to understand where the bottlenecks are. Adjusting training programmes based on what’s happening in practice, not what looked good in the planning deck.

The organisations that get AI adoption right won’t be the ones with the flashiest training programmes. They’ll be the ones who recognised that middle managers needed support and provided it before burnout set in.