A few weeks ago, a C-suite technology leader at one of our client companies told me, in a one-on-one call, that they didn’t know how to build a basic scraping tool using AI. Not a complex integration or a proprietary model, but something relatively simple for a leader who is ostensibly responsible for making core company decisions about the use of AI across the greater organization.
This isn’t a story about one person being underskilled in AI. It’s a story about what happens when a company assumes AI adoption is happening safely just because it’s happening at all, without really investing into AI education at every level of the business.
Your Team Is Already in the Wild West
Here’s what’s actually going on inside most healthcare organizations right now: employees at every level have become AI users to some degree. They’ve found tools they like, tools that make their work faster, tools nobody on the leadership team approved or even knows exist. They’re not doing this maliciously; they’re doing it because the tools work, and because nobody’s told them not to.
Think about what that actually looks like day-to-day. A marketing coordinator pastes a client’s competitive positioning into a free AI tool to speed up a slide deck. A clinical ops manager uploads a spreadsheet of patient volume data into a chatbot to “help summarize trends” for a leadership update. A junior analyst runs a contract through an AI tool to get a quick read on terms before it goes to legal. None of these people think they’re doing anything wrong. In most cases, they’re just trying to move faster and do good work. But every one of those moments is a decision about where sensitive data goes, made by someone with zero visibility into how that tool stores, trains on, or reuses what it’s given.
The problem is what happens next. If someone shares proprietary information, intellectual property, or, worse, patient data with an unapproved AI tool, that data doesn’t stay contained to the task at hand. It’s a real, tangible liability, and most companies don’t know it’s happening because nobody’s asked the question. You can’t audit what you don’t know exists, and right now, most leadership teams have no inventory of the AI tools already embedded in their day-to-day operations.
Managers Can’t Vet What They Don’t Understand
Layer on top of that a management team that, in many cases, doesn’t understand AI at all. Not because they’re incapable, because nobody’s invested in getting them there.
That creates a specific, dangerous gap: workers are producing output using AI, and managers don’t have the fluency to evaluate whether that output is any good. AI hallucinates. It gets things confidently wrong. There are patterns you can watch for, but only if you know which tools your team is using and understand enough about how those tools work to spot the red flags. A manager who can’t do that isn’t equipped to catch a hallucinated data point before it ends up in a client deliverable or a patient-facing communication.
This is why we tell every client: anything that comes out of AI needs another layer of human review. Not because the tools are bad, but because there aren’t enough guardrails yet to guarantee the output is right.
Governance Isn’t Bureaucracy, It’s the Missing Layer
Most companies skip straight to training and skip past governance entirely. That’s backwards. Training without governance just teaches people to use tools faster within a system that still has no rules.
An actual AI governance policy needs to answer three questions plainly: What tools are approved? What’s the approval process for a new one? What’s the vetting process before AI-assisted work goes out the door? Without answers to those three questions, you don’t have an AI strategy. You have a few hundred people freelancing with company data.
This Starts at the Top, Or It Doesn’t Start at All
Here’s the part leadership teams don’t want to hear: this isn’t a frontline training problem. It’s a leadership fluency problem. If your C-suite can’t speak intelligently at a high level about your AI strategy, your risks, and your governance approach, they can’t lead the people who report to them through it, no matter how good those people’s intentions are.
I want to be careful here, because this isn’t an argument against AI. AI is genuinely one of the best growth levers available to healthcare companies right now. But there’s a real push-pull in the market: some companies are rushing to adopt tools without ever building the management layer to run or vet them. Others are moving too slowly, too cautious to learn, and running the risk of falling behind entirely.
Neither extreme wins. The companies that will actually capture the upside of AI are the ones investing in leadership fluency and governance now, before the gap between what your team is doing and what your managers understand becomes the headline you didn’t want.
