The other morning I was reading The Deep View, and two numbers jumped off the screen to me. According to Deloitte’s new State of AI in the Enterprise report, 81% of companies say AI has made their decisions and execution faster. But only 25% say they’re using it to deeply transform the business. And…that 25% is down from 34% in January.

Please read that again: more AI, more speed, less transformation. Ummm…that seems wild to me!

I spend a lot of my week on the other side of this problem. At Aitherium Partners, we help leadership teams figure out where agentic workflows actually belong in their business. For instance, we have a large commercial insurance brokerage that’s growing so fast that the leaders are worried that the current sales team, made up of the most senior sales people, won’t be able to handle all of the business. Meanwhile, their business is VERY (x10) complex. So…we’re helping them rethink workflows wherein humans and AI and can work together, including tons of recent graduates, who don’t have experience in the complexities involved.

Likewise, in the DMBA program at CCA, I watch students who can now prototype a working tool in an afternoon wrestle with a much harder question: if AI can do the work faster, who decides what the work should be in the first place?

I said it before and I’ll say it again: this, my friends, is a job for designers, design leaders, and design strategists. Not because we’re the most technical people in the room (we usually aren’t), but because redesigning work is a design problem, and we’re the ones trained to solve it. Last week I wrote that designers can create the change they want to see. This is one of the biggest places that change is waiting.

Read on below…

— Justin Lokitz

Faster isn’t different

Let’s start with what the research actually says. Deloitte surveyed 3,500 senior leaders across 25 countries, and 81% told them AI improved the speed of decisions and execution. Sounds great, until you look underneath: only 7% called that improvement significant (Deloitte, 2026). The report’s own verdict is blunt: “Increased speed is no longer a differentiator; it is now the baseline.”

Here’s the line from that report I keep coming back to: “AI can shorten a process or improve a decision without changing the underlying workflow, ownership, or customer experience.”

That’s the whole problem in one sentence. Most companies took the work they already had and made it go faster. They didn’t ask whether the work should exist, who should own it, or what the customer should experience at the end of it. Jim Rowan, Deloitte’s head of AI, put it plainly to The Deep View: “The real ROI comes not only from doing existing work faster, but from redesigning work altogether.”

So who in your organization is responsible for redesigning work? I’m not talking about the person or team running it or automating it. Nope! Who’s designing it.

If the answer is “IT” or “the AI center of excellence,” you’ve found the gap.

They’re measuring the wrong thing

Wharton’s fourth annual AI adoption study, The Adaptive Advantage, found that about 40% of companies measure AI’s return in hours saved. Net-new revenue is the least-used measure, at 27%. Only 22% of managers say AI’s main job is to grow the business; 52% say it’s productivity or cost.

And yet…departments that track at least one growth measure report strongly positive returns 35% of the time. Departments tracking efficiency alone? 13%!

This is a framing failure, not a measurement failure. When the question is “how many hours can we save,” the answer will always be a faster version of today. When the question is “what could this team do that it couldn’t do before,” you get a different business.

Reframing the question is the first design move. It’s what we do in every kickoff, every discovery sprint, every time a stakeholder hands us a solution and we ask what problem it’s solving. We’ve just never pointed that skill at the operating model itself.

Nobody is looking at the same work

This is the finding that should make every design leader sit up. Wharton calls the seniority gap “the most consistent finding in this year’s data,” and puts it this way: “There is no group in this study where managers and the employees they manage describe the same working environment.”

The numbers back it up. 78% of managers say employees get adequate AI training; only 56% of entry- and mid-level employees agree. 73% of managers say workflows have changed significantly; 58% of the people doing the work say the same.

Think about what that means. Leaders are redesigning work they can’t see. They’re making decisions about roles, handoffs, and tools from a picture of the work that the people in it don’t recognize. PLEASE READ THAT AGAIN…seriously!

We have a name for that in our world. It’s designing without research. And…we’d never let a product team ship that way! (BTW, my favorite (i.e., least favorite) examples of this can be seen in many car infotainment systems, some of which seem like they were just afterthoughts thrown together from testing modules.)

This is where empathy stops being a soft skill and becomes an operating tool. The employees, managers, and agents inside a workflow are users of the operating model. They have jobs to be done, pain points, workarounds, and moments where trust breaks. Interviewing them, shadowing them, and mapping what actually happens (not what the process doc says happens) is exactly the work leadership teams are skipping.

Redesign the work, not the workflow

Deloitte’s first recommendation to leaders is literally “Redesign the work, not just the workflow.” Remove unnecessary steps, handoffs, and approvals. Move decisions closer to the work. Define where human judgment stays essential.

Most companies aren’t there…at least not yet anyway. 60% say they’ve redesigned jobs around AI to at least a moderate degree, but only 15% describe those changes as extensive. Just 15% report a significant shift in who holds decision-making authority (Deloitte, 2026).

Meanwhile, 40% already let AI execute decisions or actions without human approval to at least a moderate degree, and when AI produces a bad outcome, 23% say it’s unclear who’s accountable. Holy sh!t. That’s not a technology gap. That’s an undesigned system. (If you’ve been reading for a while, you know I’ve been banging this drum: who owns the decision now? and decision rights are the work.)

If you’ve ever built a service blueprint or even a JTBD map, you already know how to fix this. Lay out the frontstage and the backstage. Mark every handoff. Then add the layers most blueprints leave out: who decides at each step, where an agent acts, where a person approves, and who owns it when something goes wrong.

Wharton found that “autonomy is more dependent on level of risk than capability.” In other words, the question isn’t whether the AI can do the step. It’s whether it should, given what’s at stake if it gets it wrong. That’s a judgment call, and judgment is the part of design that AI didn’t compress. It’s the same muscle I wrote about in When Everyone Can Build, Taste Becomes the Advantage: knowing what good looks like, and what not to ship.

Bring people along, or don’t bother

Only 7% of the companies Wharton studied succeed on all three of its markers: proving value, transforming workflows, and aligning the organization around AI. Wharton calls them adaptive organizations, “one that rebuilds how its work gets done around AI and brings its people along with it.”

Two details from that study matter here.

First: “Departments with strong returns reorganized; they did not cut more staff than others.” The winners didn’t get there by cutting. They got there by redesigning.

Second, resistance isn’t the problem people think it is. Only 7% of respondents describe themselves as resistant, and Wharton concludes: “There appears to be no hostile segment; there is a reluctant one that is still being carried along.” At the same time, excitement about AI dropped 8 points to 51%, which Wharton calls “the first wave in which usage and sentiment move in opposite directions.”

People are using it more and liking it less. That’s a change management problem. That’s a co-design problem.

You don’t bring people along with a training deck. You bring them along by designing and co-creating with them: workshops where the team maps its own work, prototypes of new roles that people can try on before they’re permanent, and pilots framed as experiments rather than verdicts. Per my the insurance brokerage client example I started with, we didn’t just jump into an all out ideation fest about what could be, we listened and co-created a giant JTBD map with the client. We then used that map to point out areas that could be performed by AI agents, and used those to help us design a future version that could fit everyone. And…we have been prototyping and validating all of it at every step. This has been as much facilitation as design of things. Facilitation is a design power, and most AI rollouts have none of it.

Proof it can go the right way

Buried on page 19 of the Deloitte report is a small story that should be on every design leader’s wall. A furniture retailer built an AI room-design tool. You can guess what usually happens next. But the company found the tool “did not eliminate the need for human designers.” Instead, it reshaped its designers’ roles and upskilled them. The executive Deloitte spoke with described it as optimizing the workforce, not reducing it.

That’s the whole argument in miniature. AI took over part of the execution. The company redesigned the work around what people do best. And…the designers ended up doing more valuable work, not less.

So, what do we do now?

I’ve been arguing for a while here that AI is collapsing the cost of execution and pushing design leadership toward judgment and strategy. This research is what that looks like at enterprise scale. Companies have the speed. Yet what they don’t have is anyone redesigning the work, and that’s exactly what we’re trained to do!

So here’s what I’d suggest ask of you this quarter:

The companies that win with AI won’t be the ones with the most tools. They’ll be the ones that redesigned how they work and brought their people with them. That isn’t a technology job. It’s a design job. And…for once, the research is on our side.

Sources

Subscribe to Design Shift for more conversations that help creative professionals grow into strategic leaders.