The Floor Is Right. So Is the CEO. Why AI Adoption Comes Down to the CEO's Feel for AI

Doing FDE work — Forward Deployed Engineering, where we embed in a client’s operations to implement AI and business systems — there’s a scene I run into on almost every project. The people on the floor tell me one thing, and the CEO tells me the opposite.

A scene from a real meeting

We’re in a meeting about systematizing an ordering workflow. The floor lead says: “The one thing we can’t drop is this visual check. If a mistake slips through here, it lands on our customers.”

A few days later, the CEO tells me: “That check is exactly what I want to hand to AI. Keeping a good person parked on it is the biggest waste we have.”

I’m the one sitting in both meetings, so I can say this with some confidence: annoyingly, they’re both right.

“Both right” is a real thing

The floor carries today’s risk. Drop one check, cause one incident, and they’re the ones apologizing to the customer. Their caution isn’t laziness or resistance to change — it’s what responsibility looks like up close.

The CEO carries the next three years. Headcount isn’t going to magically grow, and keeping the company alive means reallocating where people’s hours go. Watching a capable person’s time disappear into a checking task and thinking “what a waste” — that’s also completely correct.

They’re just looking at different things. The floor is looking at today; the CEO is looking at three years out. Two people on different time horizons, looking at the same workflow, reaching different conclusions — that’s not dysfunction. That’s normal.

“Who’s right?” is the wrong question

Sitting between the two sides as an outside engineer, one thing becomes very clear: this isn’t a question of correctness. It’s a question of choice.

Do you accept some risk now to free up people’s time? Or do you stay on the safe side here and start automating somewhere else? The answer depends on the company: how much financial room there is, whether a mistake in this particular workflow is survivable, whether the floor has the bandwidth right now to absorb a new way of working.

And there’s an old name for the act of weighing your company’s situation and picking a direction: a management decision. It isn’t something an engineer decides, and it isn’t something a consultant’s slide deck decides. In the end, the CEO decides.

Which is why the CEO’s feel for AI matters so much

If the final call is a management decision, what determines the quality of that decision? In my experience, it’s how much hands-on feel the CEO has for AI.

CEOs who have never used AI tend to swing to one of two extremes: “AI can do all of this, right?” or “That stuff has nothing to do with our industry.” Neither of those is a judgment. They’re images.

A CEO who uses AI with their own hands, even a few times a week, carries a working sense of the boundary — this much I’d trust it with, past this point it’s still shaky. With that boundary sense, the ordering-check standoff has a resolution: “Let AI do the first pass, humans review only what it flags. Run it three months; if nothing breaks, widen the scope.” A line that honors both the floor’s concern and the CEO’s intent.

You don’t need prompt engineering skills or encyclopedic knowledge of the latest models. You need the experience of having used the thing yourself — the times it surprised you and the times it burned you. That alone upgrades a decision from an image to an informed read.

Working software changes the argument

That said, drawing lines on gut feel alone is uncomfortable. So our FDE practice is to skip the extended debate and build something small that runs on the floor.

Something remarkable happens when there’s working software in the room: abstract for-or-against turns into specifics. The floor person who said “I don’t trust AI” starts asking, “Could it read this field too?” The CEO who said “automate everything” starts saying, “Actually, a human is faster here.” The lines drawn from imagination get redrawn from evidence. Honestly, that moment might be the best part of this job.

One last thing

If you run a company and you’re weighing AI adoption, I have exactly one suggestion: use it yourself first. And if you then find that you and your team don’t see eye to eye — that’s not a sign something is wrong with your company. It’s a sign it’s healthy. The floor is right, and so are you.

How to draw the line from there — that’s a conversation we’d be glad to have with you.