A few years ago, a sales leader at my company muscled his way into owning a massive billing platform migration. He found a vendor who promised they could deliver it for $20 million in six months. He got board approval without ever consulting the people who'd actually have to build it. And one day my team and I woke up committed to a project we'd never agreed to: a project that wasn't just ambitious but genuinely impossible on the terms that had been promised.
I spent the next year in one of the worst positions a product leader can occupy: right about the problem, powerless to stop it. I pushed the team. I flagged the insanity. But the leader who'd made the commitment didn't want a pragmatic voice; he wanted compliance. It cost me a promotion and eventually my job.
That was a billing system. The pattern playing out right now with AI is the same dynamic, except it's happening in every company simultaneously. And the stakes are higher because AI touches everything.
Stop looking down. The gap is above you
We keep talking about AI adoption as a front-line problem. How do we upskill our people? What tools should we standardise on? How do we get everyone using AI?
Wrong target. The most dangerous AI knowledge gap in most organisations isn't on the product team. It's in the executive suite.
Executives are setting AI direction based on conference keynotes, podcast episodes, and board pressure, not personal experience with the technology. Many have never spent meaningful time using AI tools themselves, but they're making commitments to customers, to boards, to timelines, without the experience to evaluate whether any of it is realistic.
And product leaders are left holding the bag. One product leader put it to me bluntly: "I'm a lone voice here. Everyone's like, no, just do it. Just do AI."
Yet another product leader, a CPO, described being exhausted from "managing all these crazy, half-baked ideas that executives want to throw on my roadmap because their understanding of AI doesn't land with what the technology can and cannot do."
These aren't isolated complaints. They're a pattern. Directors quietly shielding their teams from mandates that don't make sense. Product leaders who feel like the only person in the room with actual experience of the technology everyone else is making decisions about. Meanwhile, the people closest to the work end up absorbing the gap between executive enthusiasm and operational reality.
Personal fluency isn't the fix
Leaders who've actually used the tools develop better intuition about capabilities and limitations. They generate fewer impossible requests. So, the obvious answer is: get your executives using AI. That's necessary, but insufficient. An executive who uses AI to draft emails and summarise meeting notes can still fundamentally misunderstand what it takes to make AI work across an organisation. They can still believe that buying seats equals adoption. They can still assume that individual speed automatically becomes organisational capability.
It doesn't. When individuals get faster with AI, the bottleneck doesn't disappear, it shifts. More PRDs don't mean better decisions. More competitive analyses don't mean sharper strategy. Someone has to connect the individual productivity gains into workflows that improve team outcomes and that work doesn't happen by itself.
Real organizational AI capability requires deliberate effort across at least three dimensions: engineering the right context so AI produces useful outputs for your specific domain, discovering and validating the workflows where AI genuinely adds value rather than noise, and building a learning architecture that turns individual experimentation into shared team knowledge.
Executives don't need to master these mechanics. But they need to understand this infrastructure must exist, because without it, "use more AI" is a mandate with no mechanism behind it. Without these three components, AI stays trapped in the experimentation phase and organizations never unlock the value their executives are expecting.
What you can actually do about it
I handled the billing system situation poorly. Being the person who says "this is impossible" without offering an alternative path forward just gets you labeled as a blocker. Right diagnosis, wrong approach. Here's what I'd do differently now.
Educate up, not just down. If your organization is investing in AI training for teams, push to include the executive layer, and not just tool training. Executives need exposure to what organizational AI adoption actually requires.
The CPOs I mentioned earlier? They're solving this by bringing in outside help specifically for their executive committees. Once leadership understands the difference between individual tool use and organizational capability, the quality of what they're asking for changes almost overnight.
Surface the gap. Most executives have no idea what AI adoption actually looks like inside their own organization. They see the seat licenses. They don't see that most people aren't using them, and the ones who are aren't sharing what they've learned. Make that visible. It's much harder to mandate "just do AI" when the data shows that the real problem isn't access or enthusiasm. It's that nothing connects individual effort to team outcomes.
Lead with questions, not objections. This is the lesson the billing system taught me. Instead of "that won't work," ask: "What problem does this solve, and how will we know it worked?". That one question forces strategic thinking without creating conflict. It positions you as the person driving clarity, not the person blocking progress. Being right about a problem doesn't matter if you can't bring people along.
The position only you can fill
Product leaders already sit at the intersection of business, technology, and user needs. You already translate between stakeholders who see the world differently. The AI adoption gap between executives and teams is the same work, applied to the most urgent organizational challenge most companies face right now.
It's not comfortable. It can be professionally risky—I have the scars to prove it. But no one else in the organization sees the full picture the way you do: what AI can actually do, what your team actually needs, and what your executives are missing. Closing that gap from the middle isn't just part of the job. Right now, it might be the most important part.
