The gap in most AI transformation efforts isn’t employee resistance It’s the lack of a companywide strategy that tells people what to do, in their specific role, with this specific technology.
When PwC’s US CEO Paul Griggs announced that any partner who thinks they can “opt out” of AI “won’t be here that long,” it made headlines. It’s easy to understand his sense of urgency. PwC is fundamentally restructuring its business model.
They’re shifting away from hourly billing, launching an AI platform (PwC One) that delivers some services without a an actual person involved, and shifting their hiring toward engineers and data specialists rather than traditional accountants.
The strategic direction makes total and complete sense. But there’s a ginormous execution gap between issuing an AI mandate and actually leading an AI transformation.
That’s where most companies, not just PwC, are hitting a quagmire.
What Is AI adoption strategy? And why don’t most companies have one?
An AI adoption strategy is a specific, role-level plan that defines where AI fits in an company’s operations, what employees should be doing with it, and what systems are in place to build that capability.
Most companies don’t have one. They have an AI ambition (which is really just a broad stroke order to “be AI-first”) without the operational architecture to back it up.
Telling employees to embrace AI without defining what that means for their specific role is nothing more than posturing and playing to the popularity of the headiness of the technology.
And, it paves the way for one of two outcomes…
First is paralysis. People are so freaked out by AI that they avoid the tools altogether. They don’t know when, where, and how to use them correctly so they default to not using them at all. And that puts the brakes on any transformation momentum.
The second is surface-level compliance. People use AI to check a box, but it doesn’t actually change how they work. For example, one of the most in demand but hardest to find skills is creativity. If they weren’t creative before, checking a box to report using AI isn’t going to make a bit of difference. You’ve just implemented more tech but by people who are no more willing or equipped to use critical thinking skills about if what they’re doing needs being done at all.
Neither one moves the company forward.
The distinction matters: a mandate communicates urgency; a strategy creates capability.
CEOs who confuse the two are solving the wrong problem.
Is AI adoption a culture problem or an organizational design problem?
All the struggles that come with adopting AI usually hinges on how an organization is designed, and less about culture and, how people think about it, or their willingness to try it. Framing it as any of the latter puts the burden on the people with the least amount of agency or authority to figure out a problem that when you get right down to it, is actually structural.
The companies that actually see an impact from implementing AI aren’t necessarily the ones with the most tech-enthusiastic employees or the most forceful leadership. They’re the ones that have done the boring, methodical work of…
- Mapping existing workflows to identify where AI creates genuine leverage
- Redesigning roles to reflect what humans should own versus what AI should handle
- Building operational systems (training, guardrails, feedback loops) that make the new way of working truly sustainable
This work isn’t the least bit glamorous. It’s cross-functional, slow, and always means you have to connect the dots between strategy, HR, technology, and the people who do the actual work.
But it’s also what separates organizations that are genuinely transforming from those that’re amping up internal pressure without producing any real results. Cases in point…
USAA
The company introduced an AI assistant to help 16,000 customer service workers provide correct answers to member questions. While employee feedback was positive, the company reported having no calculation of financial payoff or ROI for the software as of late 2025.
JPMorgan Chase
Despite having 200,000 staff using a secure AI assistant, the bank has shut down hundreds of AI projects that failed to move past the experimentation phase. While some individual tasks are faster, enterprise-level productivity gains remain difficult to quantify.
Johnson Controls
The company rolled out an AI tool to save 10–15 minutes per repair call, but as of August 2025, it remained in a limited rollout with no clear enterprise-wide ROI yet.
A CEO’s responsibilities in an AI transformation
CEOs and their leaders have three core responsibilities in an AI transformation that go beyond setting strategic direction…
1. Define specifically where AI fits in the business strategy.
Not “AI is a priority,” but: Which functions? Which workflows? Which decisions should AI inform, accelerate, or own? And which ones require human judgment at the center? This means taking a hardcore look at the operating model and making deliberate trade-offs, not just issuing broad mandates.
2. Prepare people for how to use AI in their roles.
This is the step most companies skip. If AI is absorbing the first layer of tax analysis or M&A due diligence, like PwC is building toward, the person who used to do that work needs to understand what their role and work look like now. Their job doesn’t disappear, it transforms. But people can’t reinvent their roles in a vacuum, it would be complete and utter harry carry. Leadership across the board has to define the new scope of human judgment and provide a credible path for employees to develop the skills required to operate there.
3. Build the infrastructure that makes adoption possible.
This includes role-specific training (not just generic AI literacy modules), clear policies on what’s appropriate use and what’s not, time and resources to experiment (and yes, accept some level of failure along the way), and a culture where asking “I’m not sure how to use this” is safe rather than career-limiting.
What are the risks of an AI mandate without a supporting strategy?
Organizations that issue AI mandates without also developing a thorough strategy that supports it face three immediate risks: surface-level compliance, increased operational risk, and attrition of high-value employees.
Surface-level compliance happens when people use AI tools to satisfy expectations rather than change how work gets done. This leads to checking all the right boxes so adoption metrics look good. But the company never actually gets any better using it.
Operational risk goes up when employees use AI in contexts that it wasn’t designed for, or trust what it delivers without taking a good look at the output. It’s not because people are careless. It’s because they were never given clear guidance on where AI judgment is reliable and where human review is absolutely essential.
Attrition happens when people look at the company’s AI direction, can’t see a coherent future for their skills, and they leave. Worse yet is that these are often the most thoughtful, experienced employees.
And it’s the attrition that rarely shows up in the headlines about “AI resisters.”
Ultimatums drive urgency. They don’t build capability.
What effective AI leadership looks like
Effective AI leadership pairs urgency with clarity of what the future looks like for individual employees and the company as a whole. A concrete vision of what work looks like after AI is integrated, and they’ve made a real investment in helping people get there.
This means…
- Communicating that AI is a priority along with what it means for specific roles and teams
- Creating role-specific learning pathways, not one-size-fits-all AI training
- Measuring whether the company is becoming more capable, not just whether tools are being used
- Senior leaders modeling AI-integrated work themselves, not just mandating it from above
The pace of AI change is real, and the window for competitive advantage is small. CEOs who communicate that urgency are not wrong. But urgency without a road map is just pressure — and pressure alone does not produce the organizational transformation that AI actually requires.
The bottom line on AI adoption strategy
The organizations that win the AI transition won’t be the ones that issued the most forceful mandates. They’ll be the ones that did the harder work. They figured out where AI fits in their strategy, defined what their people should do with it, and built the systems to make that possible.
Telling people to adapt is the easy part of AI but it’s not true leadership. Building an organization where they actually can is the hard part. And the most consequential leadership challenge of the next decade.
