In the early days of the Toyota Production System (TPS), Taiichi Ohno, the father of modern manufacturing, used to practice a radical form of management training. He would take a manager down to the factory floor, draw a circle on the ground with a piece of chalk, and tell them to stand inside it.
The instruction was simple: Observe. The manager wasn’t allowed to leave the circle until they had identified the “muda” (the waste) clogging the production line. This was genchi genbutsu (going to the source) in its purest form. It wasn’t about doing the work; it was about truly seeing the work to ensure the system functioned as intended.
A slightly different perspective from Shigeru Ohsawa, metalworker on the actual Toyota production line at the time of Taiichi Ohno is particularly instructive here:
With the introduction of the production line, the foreman’s job changed. His role now was to see that the line went smoothly. Before that, they thought that wasn’t enough, that if they weren’t getting their hands dirty they weren’t really working. But Mr. Ohno would come along and say ‘your job is to stand here and make absolutely sure that the line is moving properly’. I remember him drawing a chalk circle on the floor and saying ‘you stand in here’.”
Today, as I navigate the rapid evolution of Generative AI in software development, I find myself standing in a new kind of chalk circle.
The Temptation of the IDE#
Lately, many of my peers in the fintech and tech space keep asking me: “why aren’t you using Cursor? Why aren’t you using Anti-Gravity’ and these AI enabled IDEs that essentially write the code for you while you type?”
They aren’t wrong to ask. These tools are incredible. They are the high-tech equivalent of the automated power tools that replaced the manual hammer. But I’ve been militant about a different approach: I use Generative AI models almost exclusively through the Command Line Interface (CLI). In the Terminal.
Why? Because for a leader, the terminal is your chalk circle.
Promoting Yourself to Foreman#
When you work inside a modern AI-integrated IDE, the friction between thought and code is so low that you are constantly tempted to “get your hands dirty.” You find yourself tweaking syntax, fixing indentations, and wrestling with the “metal” of the codebase. You become a worker again.
By forcing yourself to interact with AI through the terminal, you maintain a deliberate, strategic distance. You are not “coding” in the traditional sense; you are managing a production line of intelligent agents.
In the era of AI, the biggest mindset shift isn’t learning how to prompt, it’s realizing you have been promoted. You are no longer the craftsman at the bench; you are the foreman standing in the Ohno Circle. Your job is to make sure the agents are equipped, the logic is sound, and the line keeps moving.
The New “Muda” (Waste) in AI Coding#
In the Toyota plant, waste was overproduction or defective parts. In AI-enabled coding, waste looks different:
- Hallucinated Logic: Code that looks right but fails at scale.
- Feature Creep: The AI building things you didn’t ask for because you weren’t watching the “line” closely enough.
- Technical Debt: Quick fixes that satisfy the immediate prompt but break the ecosystem.
When I stand in the “chalk circle” of the terminal, I am forced to observe the output of the model as a supervisor, not a peer. I ask questions. I identify where the “line” is snagging. I am practicing genchi genbutsu at the architectural level rather than the character level.
Scaling the Venture, Not Just the Code#
Throughout my career, from founding UBX to scaling corporate venture funds, the lesson has always been the same: Systems beat individual effort every time. If you spend your day with your hands in the grease, you lose the situational awareness required to win the larger battle. Taiichi Ohno applied the Toyota Production System to the world of car manufacturing; now, AI is doing the same for the world of digital creation.
If you want to lead in this new era, stop trying to be the fastest typist in the room. Draw your circle. Step inside. Use the terminal to observe the flow, eliminate the waste, and make sure the line is moving properly.
The era of the “coder” is transitioning into the era of the “system architect.” Make sure you’re standing in the right spot to see the difference.
Read Next
