A support rep at a mid-size software company used to spend twenty minutes drafting a reply to an escalated ticket. She uses AI now. The draft takes four minutes.
She still routes that ticket through the same three approval steps, built for a process that moved at the speed of a twenty-minute draft. The reply got faster. The path around her did not move at all.
Multiply that across every seat in the company, and you get the current shape of AI adoption. Wide use. Barely-moved outcomes.
The Metric Measures the Wrong Thing
Publicis Sapient asked companies this year whether AI touches most of their work. Seventy-three percent said yes. Then they asked whether AI is core to how the business actually runs. Ten percent said yes to that one.
Near-universal usage sits right next to almost no structural change, and most people in charge already know which of those two numbers ends up on the board slide. Adoption rate is clean. It moves in one direction. It looks good next to a competitor’s slower number.
Business change doesn’t photograph well at all. It looks like a decision that used to need three approvals now needing one. It looks like a role that used to own a handoff quietly not owning it anymore. That kind of change is slow and uneven and doesn’t fit in a single percentage point, which is exactly why it keeps losing to the number that does.
What Redesign Actually Requires
A faster draft inside an unchanged process is not redesign. Redesign means someone decides that a step, a handoff, or an entire approval chain no longer needs to exist in its current form. That decision reassigns ownership. Someone who used to review no longer does. Someone who used to wait now doesn’t either.
Forty-two percent of those same respondents said flat out that their own organization isn’t built to capture the value AI already makes possible. Not a capability gap. The capability’s sitting right there, already paid for, already working in the demo. It’s a decision gap. Somebody with the authority to remove a step just hasn’t removed it yet.
This Isn’t a New Problem
This is the digital transformation conversation most companies never actually finished. A decade ago, moving a paper form into a web portal got called transformation. It wasn’t. It was digitization, the same process wearing a browser. The real work, rebuilding the process instead of just relocating it, got skipped because the digitized version already looked like progress on a slide.
AI doesn’t tolerate that shortcut the same way a portal did. It sits directly on top of whatever process already exists. A process that skipped its redesign the first time around just gets a faster version of the same unfinished job. The tool doesn’t hide the debt anymore. It exposes it, in real time, on every ticket, every approval, every handoff.
That’s the actual requirement underneath all of this. Not “adopt AI.” Finish the transformation the business owes itself, the one AI is now forcing into the open.
Where the Resistance Actually Sits
Teams pick up a faster way to do their existing job without much of a fight. Nobody turns down a tool that makes their own work quicker. Ask them to accept a redesign that changes the shape of the job entirely, and you get a different reaction.
The resistance people in charge keep running into was never about AI. It’s about giving up territory. A faster draft doesn’t threaten anyone. Cutting three approvals down to one does, because that cut decides who still matters at that point in the workflow, and who doesn’t.
That’s a people problem dressed up as a technology problem. Organizations respond to it with a training rollout and a communication plan, because those are the tools built for adoption problems. What’s actually on the table is a decision about authority, and most people would rather run another training session than have that conversation out loud.
The Pattern Has a Name
Call it adoption theater. Usage numbers climb, the operating model doesn’t move, and the gap between what the tool can do and what the business is actually capturing just keeps getting wider. The tool gets faster every quarter. The org chart looks exactly like it did the day before anyone opened the tool for the first time.
Nobody closes that gap by accident. Faster tools bolted onto the same structure just produce faster versions of the same old output. Leaders who confuse that with progress will spend another year putting up adoption numbers while the actual business underneath stays exactly where it was.
Deploying a tool is not the hard part. It was always the digital transformation that the business skipped the first time; now AI is compressing time to make that transformation non-negotiable.