There’s a curious disconnect emerging in conversations about AI in the workplace, particularly for the management support profession. Over the past several months, I’ve spoken with dozens of executive assistants and administrative professionals about their experiences with AI implementation. What they told me contradicts much of the prevailing narrative about rapid technological transformation.

This reality was recently confirmed by surveys on both sides of the Atlantic. Research in the Netherlands found people are using AI more at home than at work, with employers harboring serious doubts about implementing AI tools due to data protection and privacy laws. There’s also a widespread perception that AI tools can make mistakes that risk reputation and financial damage.

Even more striking, a March 2025 Pew Research Center survey of over 5,000 U.S. workers revealed that 81% are considered non-AI users, saying little or none of their work is done with AI. A full 17% of workers sampled hadn’t even heard of AI being used in their workplace. Contrary to predictions of rapid workplace transformation following ChatGPT’s launch in 2023, the AI revolution remains largely theoretical for most workers two years later.

“I went to this workshop about bringing AI tools into our workflows,” one senior EA at a global financial services firm told me recently. “It was fascinating, but when I tried to actually implement what I’d learned, our IT security protocols wouldn’t allow it. I spent weeks trying to get approval and eventually gave up.”

This isn’t an isolated experience. It’s becoming the norm.

The implementation chasm

I observe a significant gap between theoretical possibilities of AI and practical realities of organisational life. While technology evangelists and trainers enthusiastically promote comprehensive AI-powered productivity tools, the actual adoption pattern looks very different on the office floor.

Many management support professionals are caught in a frustrating paradox: they’re simultaneously told that mastering AI tools is essential for their professional survival while encountering substantial organisational barriers to actually using those tools.

I recently facilitated a roundtable discussion with twelve executive assistants from various industries. When I asked how many had successfully implemented comprehensive AI tools in their daily work, only two raised their hands. The others described a patchwork approach, using personal AI assistants for specific tasks while navigating complex approval processes for more integrated solutions.

“My executive keeps sending me articles about how AI is going to transform my role,” one participant shared. “Then in the next breath, he tells me our company won’t approve the tools I’ve requested because of data security concerns. It’s maddening.”

The organisational reality

These conversations reveal something that should be obvious but is rarely acknowledged in discussions about workplace technology: adoption isn’t primarily about the tools or even the training to use them. It’s about governance structures, security protocols, legacy systems, budget allocation processes and the invisible power dynamics that determine who gets access to what.

These organisational realities haven’t fundamentally changed in decades, despite the revolutionary promises of each new technology wave. I witnessed the same patterns introducing customer relationship management systems in the early 2000s, collaboration platforms in the 2010s and now AI tools in the 2020s.

The cycle is depressingly predictable:

Technology providers sell directly to enthusiastic executives who envision transformative possibilities. Initial pilots show promising results. Then implementation falls to support staff who immediately encounter the practical barriers of organisational life, from incompatible legacy systems to unclear data governance policies.

The resulting adoption is typically fragmented and uneven, creating additional complexity rather than the promised efficiency.

The pivot to personal assistants

What I’m now observing is a pragmatic pivot. Both management support professionals and those training them are increasingly focusing on personal AI assistants that can be used within existing constraints rather than comprehensive solutions requiring extensive organisational approval.

This approach makes practical sense. If you’re unlikely to get access to integrated AI tools anytime soon, why invest time learning them? Better to focus on capabilities you can actually implement. Yet this creates precisely the situation revealed in the Dutch survey. People use AI tools more in their personal lives than in professional contexts, because the organisational barriers remain unsurmountable.

But this pragmatic adaptation comes with its own risks and challenges for AI in the workplace. Personal AI assistants exist outside organisational workflows. They create parallel processes rather than integrated solutions.

And they potentially introduce new security and governance concerns that may not be immediately visible, precisely the data protection and privacy issues that organisations are worried about in the first place.

There’s a certain irony in organisations blocking official AI tool adoption only to inadvertently push employees toward personal AI solutions that may actually increase rather than mitigate the very risks they’re concerned about.

The capability that truly matters

This messy reality of AI in the workplace reveals something I’ve been arguing for years: the critical capability for management support professionals isn’t tool mastery but systems thinking.

The most valuable administrative professionals aren’t those who can use specific AI tools most efficiently. They’re those who understand how organisations actually function beneath their formal structures. This is the very thing that management support professionals are uniquely positioned to navigate.

When I worked with a particularly effective executive assistant at a media company years ago, what distinguished her approach wasn’t her technical skills but her ability to see patterns across seemingly disconnected parts of the organisation.

She understood which teams could collaborate effectively, where information flowed smoothly, and where it got stuck. This systems understanding allowed her to bypass formal processes when necessary while strengthening them when possible.

That capability is even more valuable in today’s fragmented technological landscape.

Beyond the tool-mastery trap

The current focus on mastering specific AI tools creates a dangerous illusion that technological proficiency alone will secure professional relevance in the AI in the workplace revolution. The reality is far more complex.

The future of AI in the workplace doesn’t belong to those who can prompt ChatGPT most effectively or build the most sophisticated workflows in Notion. It belongs to those who understand the complex human systems in which those tools must operate and can navigate the gap between technological possibility and organisational reality.

This isn’t about resisting AI adoption. It’s about recognising that implementation happens within existing organisational contexts that can’t be ignored or wished away.

The Pew study reveals that even those using AI are primarily deploying it for simple tasks: 57% for information seeking (Ofcom found last month that in the UK 32% have replaced Google and other search engines with AI), 52% for editing and 47% for drafting content.

Only a small minority are leveraging AI for more complex tasks like ideation (35%) or analysis and coding support (27%). This suggests that even as AI capabilities advance, organisational constraints and user adaptation remain significant limiting factors.

The management support professionals who thrive in this environment will be those who can:

• Identify where AI tools can realistically be integrated given existing constraints
• Navigate approval processes by understanding both formal procedures and informal influence
• Bridge the gap between enthusiastic executives and cautious IT departments
• Recognise when workarounds are necessary and when they create more problems than they solve
• Communicate effectively about both possibilities and limitations

These aren’t technical skills. They’re systems capabilities that come from understanding how organisations actually function rather than how they appear on the organisational chart.

The honest conversation we need

What management support professionals need right now isn’t another workshop on prompting techniques or another webinar about AI-powered email management. Those have their place, but they’re insufficient.

What’s needed is an honest conversation about the messy reality of AI in the workplace and technological adoption in complex organisations. A conversation that acknowledges the gap between theoretical possibility and practical implementation. A conversation that values systems understanding as much as tool proficiency.

The AI revolution will eventually transform the management support profession and redefine AI in the workplace, but the path won’t be as direct or as rapid as many predict.

When Pew Research finds that even two years after ChatGPT’s launch, 81% of U.S. workers still don’t use AI in their work, we need to acknowledge that workplace transformation moves at institutional, not technological pace. This reality will be shaped by the stubborn realities of organisational life that haven’t fundamentally changed despite decades of technological innovation.

Understanding that reality – and developing the capabilities to navigate it – is what will truly future-proof the management support profession.