In 1948, Taiichi Ohno, an engineer at Toyota, went and stood on the factory floor. Literally. He drew a chalk circle on the ground and stood inside it. For hours. Watching how the work was actually done. Not how it was described in the manual. How it really went.
What he saw was revealing. Workers searching for materials that were not in the right place. Waiting times between steps that nobody had noticed. Actions that once had a reason but had long since lost their purpose. The gap between how the work was designed and how it was actually done was enormous.
From those observations grew the Toyota Production System. The foundation of lean manufacturing. A revolution that began with a chalk circle and the patience to watch.
The meeting room trap
AI strategy is almost always devised in a meeting room. A team of managers brainstorms about possibilities. Roadmaps are drawn. Vendors are invited. Business cases are written.
The problem: nobody has gone to look.
The people who do the work are not in that meeting room. The processes that cause daily frustration are not in the process handbook. The workarounds employees have invented to make the system bearable are invisible to management. Sometimes they are already using AI tools on their own initiative just to keep the work manageable. That phenomenon, Shadow AI, is one of the most reliable signals that friction exists.
And so the AI strategy ends up targeting the wrong things. Processes that look important on paper. Problems that managers see but employees do not recognise. Solutions to symptoms while the cause sits somewhere else entirely.
The chalk circle method
Go and sit next to someone. Twenty minutes. Watch while they work. Let them explain what they are doing and why.
Three questions are enough.
"What is frustrating about this?" People know exactly where the friction is. They have just never told anyone, because nobody asked.
"Why do you do it this way?" The answers reveal the history of the process. "Because the system won't accept it otherwise." "Because my predecessor did it this way." "Because we once made a mistake and this control step was added." Layers and layers of workarounds, accumulated over years.
"What would you do if this step didn't exist?" This is the question that opens things up. It lets people think about the work without the constraints of the current system. This is where the best ideas come from.
Where friction is, value is
Friction in a process is a signal. It tells you where energy is leaking. Where people spend time on work that adds no value. Where systems are not doing what they should.
The places with the most friction are almost always the places where AI has the most impact. Because friction makes visible what can be automated, simplified or eliminated.
But there is a catch. Sometimes friction is a symptom of a broken process. And accelerating a broken process with AI only makes it break faster. This is the core of solutioneering: attaching a tool to something that first needs to be redesigned. So the first question for any friction is: should this process exist at all? Is it valuable, or just old?
If the answer is "old," don't automate it. Retire it.
From observation to action
The step from watching to doing is smaller than you think. Choose the friction that comes up most often. The smallest piece with the most pain. Build something for it in two weeks. Test it with the people you observed. They are your best judges, they feel immediately whether it makes their work better.
This is how the best AI automation projects begin. Not with a roadmap, a business case or a vendor selection. With twenty minutes of watching. With the willingness to be surprised by what you see.
Ohno sometimes stood in his chalk circle for six hours. You do not need to go that far. Twenty minutes is enough to see what a month of meetings cannot produce.
The best AI strategy starts by walking out the door and going to see how the work is really done. Everything else follows naturally.