In the early 1950s, a Cincinnati company was quietly going under. It made a soft, putty-like compound that cleaned coal soot off wallpaper. Then two things happened: homes switched to gas heat, and vinyl wallpaper arrived. The soot disappeared, and so did the customers.

The product still worked, but the market for it was gone.

What saved the company wasn't a better formula. It was a nursery school teacher who noticed her students loved squishing the stuff. The company added color, dropped the cleaning agents, and renamed it Play-Doh. Same compound, new direction. Three billion cans later, it sits in the National Toy Hall of Fame.

Most companies are sitting on their own version of Play-Doh right now. AI is changing what their people spend their days on. The reflex is to ask, "what can we cut?" The better question is the one the Play-Doh team stumbled into by accident: where does the value come from now? 

Most AI conversations start with subtraction: which roles go away, how many hours we can save, and what we can stop paying for.

Those impacts can be real, but they’re also small.

The companies pulling ahead ask additive questions instead. Now that AI handles the routine, what work finally becomes possible? What did our people never have time for? Where does their judgment matter most?

Answering those questions requires two things: a clear destination and a way to measure progress toward it.

Most reorgs start with today and try to optimize it. They rearrange people around yesterday's work, then leaders wonder why nothing changes.

Start with the future instead, and work backward to now.

• Define the future state first. What does the team need to be capable of in 18 to 24 months? Not today's work, faster.

• Mark the milestones in reverse. Work back to the present, noting the turning points along the way.

• Find the gaps. What skills, roles, or capabilities are missing to make that future real?

• Build around closing them. Roles and structure follow the gaps, not the other way around.

There's research behind this. Park, Lu, and Hedgcock found in Psychological Science that people who plan backward from a goal outperform those who plan forward. More motivation and less time pressure drive better results. The effect was strongest for complex goals.

A destination sets direction. People still need a target to know if they’re on track. Before you reskill anyone, define the outcome across three dimensions:

• Quality. What does the work need to be and do? Example outcomes include on-brand, accurate, or accessible.

• Impact. What improves for the customer and the business? That might include satisfaction scores, revenue, or customer lifetime value.

• Reliability. How well and how fast does it happen? Resolution time, review hours, or first-pass approval rates could reflect success in this area.

Those answers become your KPIs that tell a person whether they're winning. That’s the difference between a team that uses AI with purpose and one that just stays busy with it.

IKEA put a chatbot named Billie in front of customer service. It now resolves around half of inquiries on its own. The easy move would have been layoffs.

Instead, IKEA studied the leftover issues that Billie couldn't address: people who wanted help with room layout and design before they bought. That was a business no one had staffed for.

So IKEA reskilled 8,500 service reps into interior design consultants. They built on what those people were already good at—listening, guiding, earning trust—and added design skill on top. The result was a new service worth roughly €1.3 billion a year.

Same people. New direction.

The Federal Reserve Bank of St. Louis estimates that AI hands back nearly one full workday a month for a regular user. That’s a good thing when it’s put to use like IKEA did. It’s a waste when people fill those hours on reflex: more tickets, more decks, more email.

That seems productive, but it’s not. It just makes us feel productive compared to spending time deeply thinking. 

Research in Communications of the ACM found that heavy AI users report their work feels easier, while their capacity for independent judgment erodes. The organization gets faster and flatter at the same time, creating more output that’s less interesting.

Instead, spend the recovered time on what AI can't do. Run small experiments where the cost of being wrong is low. Share what you learn and make room to ask whether you're still working on the right problems.

The pattern holds across decades and industries.