Imagine handing a master carpenter a brand-new power saw — and changing nothing else about the shop.
Same bench, jammed against the same wall. Same cramped aisle. The same rule that says ask before you cut. You've handed her a faster tool and left her standing in a workflow designed for a hand saw. Within a week, the saw is either gathering dust or, worse, throwing sawdust into work that used to be careful. You didn't make her better. You bolted horsepower onto a job that was never redesigned to hold it.
That is exactly what most organizations are doing with artificial intelligence right now. We are buying extraordinary tools and dropping them onto jobs that were drawn up for a slower world. I call it the bolt-on trap, and it is the single most common reason AI deployments quietly disappoint — not the model, not the vendor, the job. When you bolt AI onto an unchanged role, you don't remove friction. You add a new thing to learn on top of all the old friction, and you hand your best people one more reason to wonder whether the tool is here to help them or replace them.
I have spent nearly four decades inside organizations celebrated for how their people perform — Starbucks, The Ritz-Carlton, Mercedes-Benz, Airbnb, Zappos, and most recently One Medical. The lesson that never bends is this: technology only lifts people when the job around it is redesigned to let it. The tool is the easy part. The work is the work.
The adoption gap nobody is watching
Let me give you the data, because the case for redesign is not a hunch. It is measurable.
In June 2025, Gallup reported that AI use at work in the United States has nearly doubled in two years — from 21% of employees using it at least occasionally in 2023 to 40% in 2025. Frequent use roughly doubled too, from 11% to 19%. On paper, that looks like a revolution.
Now look underneath it. That same Gallup research found that frequent AI use among production and frontline workers has stayed essentially flat — 11% in 2023, 9% in 2025. The revolution is real, but it is happening almost entirely in white-collar roles, where 27% now use AI frequently. The people who are your customer experience — the ones at the counter, on the phone, in the clinic, on the floor — have barely been touched by it.
That is not a technology gap. It is a job-design gap. Frontline roles weren't redesigned to make room for these tools, so the tools never arrived in any way that mattered.
And here is the number that should stop a leadership team cold. Among employees whose organizations have started rolling out AI, only 16% strongly agree the AI tools provided are useful for their work. Read that again. We are deploying at scale, and five out of six people can't honestly say the tool helps them do their job. We bought the saw. We never rebuilt the shop.
Why the bolt-on trap manufactures fear
Here's the part the spreadsheet misses. When you add a tool without redesigning the job, you don't just lose productivity. You generate anxiety — because an unredesigned job leaves the meaning of the tool ambiguous, and people fill ambiguity with their worst fear.
Gallup found that 44% of employees say their organization has begun integrating AI, but only 22% say it has communicated a clear plan for doing so. That 22-point chasm is what I'd call the clarity deficit — the gap between what is happening to people's jobs and what anyone has bothered to explain. In that vacuum, a new tool doesn't read as an upgrade. It reads as a question: Is this here to help me, or to replace me?
The data says clarity is the antidote, not the slogan. When employees strongly agree leadership has communicated a clear plan for integrating AI, Gallup found they are three times as likely to feel very prepared to work with it and 2.6 times as likely to feel comfortable using it. The same tool, the same week — but landing on a redesigned, clearly-explained job instead of a silent bolt-on. The facts are identical. The fear is not.
What "redesign the job" means
Redesigning work for AI is not a poster about innovation. It is a sequence of unglamorous decisions, and it starts with one question I'd put at the top of every rollout: What friction should this tool remove from this specific job — and what human value does that free the person to add?
Notice the question has two halves, and most organizations only ask the first. Subtract the friction. Map the role honestly and find the drudgery — the copy-paste, the manual lookup, the form that exists only to feed another form. That is where AI earns its keep, and it's exactly the work your people have been begging you to take away.
But subtraction alone just produces an emptier job, and an emptier job is a frightening one. So you have to ask the second half: and now, raise the human value. What does the person do with the time and attention you just gave back? In a redesigned role, AI absorbs the transactional layer so the human can move up — to judgment, to the difficult conversation, to the unrepeatable moment that turns a satisfied customer into a loyal one. Redesign moves people up the value chain. Bolt-on just makes them anxious operators of one more screen.
There is a virtuous loop hiding in this, and Gallup measured it. The same research found that 68% of employees who have personally used AI to support customer interactions say it had a positive effect — compared with only 13% of those who haven't tried it. Experience converts skeptics. But experience only happens when the job is designed to give people a real, friction-removing reason to pick the tool up in the first place. Redesign creates the experience; the experience dissolves the fear.
Three moves to redesign the job, not just buy the tool
1. Map friction before you map features. Before you evaluate a single AI vendor, sit with the people in the role and list the parts of the job they'd happily never do again. Those are your targets. A tool chosen to kill named, hated friction gets adopted. A tool chosen because it demos well gets ignored — and resented.
2. Redesign the role on paper, out loud, with the people in it. Don't hand down a finished job. Show your team the redesign: here's the drudgery the tool removes, here's the higher-value work it frees you for, here's why we need more of you, not less. That conversation is the redesign. It closes the clarity deficit and converts the tool from threat to relief.
3. Reinvest the time you just freed — visibly. The fastest way to confirm everyone's worst fear is to remove friction and then quietly pocket the hours as headcount savings. The fastest way to build engagement is to point the reclaimed time at work that is unmistakably human and unmistakably valued. Name where the time goes. Protect it on purpose.
The reframe that changes the math
Here is the shift I urge every leadership team to make. Most organizations treat AI as a tool-acquisition problem: pick the platform, push it out, count the licenses. Flip it. Treat it as a job-design problem that happens to involve a tool. The platform is the smallest decision you'll make. The job is the one that determines whether AI reduces friction or manufactures fear.
Because the carpenter never needed a faster saw standing in an unchanged shop. She needed a shop redesigned around what the new saw makes possible. Your people are no different. The tool is not the upgrade. The redesigned job is the upgrade — and it is the one thing your competitor, frantically bolting AI onto unchanged roles, is almost certainly forgetting to build.
So let me leave you with the question worth carrying into your next planning session: when AI arrives in your organization, are you redesigning the job around it — or just bolting a faster tool onto the same old bench, and calling the resulting anxiety "resistance to change"?
If that question is harder to answer than it should be, that is exactly the work I help leadership teams do. Come find me at josephmichelli.com/contact, and let's design jobs your people — and your customers — can feel the difference in.
