When Jensen Huang said he wants his engineers to “spend zero time writing code,” the headlines made it sound like technical expertise is dying. That AI will write all the code so humans aren’t needed.
I think that interpretation misses what he’s actually describing.
In a professional kitchen, the Executive Chef doesn’t cook. They stand at the pass, the stainless steel shelf where the kitchen meets the dining room, and they taste every single dish before it goes out. They orchestrate timing across stations. They catch problems before they reach the guest. Their expertise isn’t in production anymore. It’s concentrated entirely into judgment.
That’s the shift Huang is pointing toward. Not the elimination of experts, but the elevation of them. The mechanical work gets delegated. The judgment work becomes the whole job.
I learned this principle the hard way, standing in front of a wood-fired pizza oven, watching my eleventh pizza of the night turn to charcoal.
Reading the Oven
A wood-fired pizza oven is not like your oven at home. Home ovens are predictable. You set them to 425°F, wait for the beep, and trust that the temperature is what it says it is. A wood-fired oven is alive. The temperature shifts constantly depending on where the fire is burning, how recently you added wood, and even how humid the kitchen is that night. The back right corner might be 750 degrees while the front left is closer to 550. The stone floor retains heat differently than the dome. And the only way to know any of this is to watch, to listen, and to feel.
I knew none of this my first night on pizza station.
The head chef had shown me the basics during a slow Tuesday afternoon. Load the pizza onto the peel. Slide it into the oven at the right angle. Rotate it every fifteen seconds or so. Dome it to melt the cheese. Pull it when the crust looks right. Let it cool, then cut.
Simple enough when you have one pizza and no pressure. Completely different when you have six tickets hanging and you’re building while cooking at the same time.
My first pizza came out beautiful. Beginner’s luck. The second one I left in too long because I was building the next one, and the crust turned into charcoal. The third I pulled too early because I was overcorrecting, and the dough was still raw in the center. The fourth I placed in a dead spot where the stones had cooled, and it just sat there, slowly getting soggy while I wondered why it wasn’t cooking. By the eighth pizza, I was so far behind that I started rushing, which only made everything worse.
The head chef stepped in around pizza number nine. He didn’t take over the station. He stood next to me, watching the oven the way a mechanic listens to an engine.
“See how the flames are licking up the right side?” he said. “That means the heat is concentrated there. You want to start your pizza on the left, let the bottom set, then rotate it toward the flames for the final char.”
He pointed at the oven floor. “See that dark spot? That’s where you pulled the last burnt one. The stone is scorched. It’s hotter than everywhere else now. Avoid it for the next ten minutes.”
I asked him how he knew all of this without a thermometer. He shrugged. “You don’t read the temperature. You read the oven.”
Over the next few weeks, he taught me what reading the oven actually meant. The color of the flames tells you about oxygen flow. The sound of the fire tells you if you need more wood. The way the cheese bubbles tells you if the heat is coming from above or below.
It was accumulated knowledge, built through thousands of hours of watching and adjusting and making mistakes.
Eventually, I stopped burning pizzas. Not because I memorized a set of rules, but because I developed an intuition for how the oven behaved. I could glance at the fire and know where to place the next pie. I could hear when the wood was getting low. I could feel when the stones needed time to recover.
That intuition is what the head chef had that I lacked. And it’s what allowed him to step back from the oven entirely and manage the entire kitchen instead. I was barely able to focus on the oven and get it right, he had to watch three other stations and make sure that hundreds of dishes went out every hour without a major mistake.
The Pass
The Executive Chef stands at the pass during service. They’re not flipping burgers or whisking sauces. They’re watching every station simultaneously. They’re timing service so that the protein and the vegetable and the starch all land on the plate at the exact same second. They’re calling out “firing table seven” so the grill cook knows to start the steak, and three minutes later calling “walking in the sides” so the vegetable station starts plating.
Most importantly, the Chef is the only person who touches every single plate before it goes out. They pick up a spoon, taste the sauce, and decide if that dish is worthy of the restaurant’s name.
To do this job well, the Chef has to understand every station in the kitchen. They have to know how the oven behaves, how the grill cooks, how the sauces reduce. Not because they’re going to do all that work themselves, but because they need to recognize when something is wrong before it reaches the guest.
This is the model for working with AI tools.
When I was burning pizzas, I was doing the manual labor of cooking. I was focused on the physical act of sliding dough into an oven. My attention was consumed by the mechanics. I didn’t have the bandwidth to think about timing, or coordination with other stations, or whether the overall flow of service was working.
The head chef, standing at the pass, had delegated all of that mechanical work. His job was to taste, to orchestrate, and to catch problems before they reached the guest. He could manage the entire kitchen because he wasn’t stuck managing a single oven.
The mechanical work of writing syntax, the boilerplate code, the repetitive logic, can increasingly be delegated to AI tools. This doesn’t eliminate the need for expertise. It concentrates it. Instead of spending eight hours writing code, an engineer might spend only two hours prompting for code and six hours reviewing, architecting, and ensuring the output actually solves the right problem.
The value of expertise shifts from production to judgment. From chopping the shallots to organizing the flow of service.
Scaling the Brigade
In a traditional kitchen, we use the Brigade System to handle complexity. Each cook owns a specific station. One person handles the grill, another makes salads, another manages the oven. The head chef doesn’t need to be actively cooking at every station.
They need to be skilled at recognizing quality and coordinating timing. They also need to be capable of covering any station at any moment either because the cook is overwhelmed or in some cases because the cook cannot handle it and quits.
The same principle applies to working with specialized AI tools. You might have one tool that handles data queries, another that manages scripting logic, another that reviews for errors. Each tool has a specific job, just like each station in a kitchen. Just like cooks, sometimes AI tools get overwhelmed and you need to do the job yourself.
Your job, as the expert, is to orchestrate. To know which tool to deploy when. To check the output before it ships. To recognize when the sauce is broken even if you weren’t the one who made it. Because you still need to be able to fix that broken sauce or function before it reaches the client.
This is how you scale without losing quality. One expert managing ten specialized tools can produce more than ten generalists working independently. But only if that expert actually understands what good output looks like. Only if they can taste the pass.
The Cost of Outsourcing Your Palate
Here’s where the trouble starts. It’s tempting to see AI as an opportunity to remove experts from the loop entirely. If the machine can write code, why pay someone who understands code?
This is like a restaurant owner buying a pizza oven and assuming they no longer need someone who understands how ovens work. When the crusts come out burnt, they won’t know why. They can’t read the oven. They’re at the mercy of a tool they don’t understand.
In AI terms, this creates what I call Mystery Meat. Code or output that works, but that nobody in the organization can actually explain. At low volume, Mystery Meat might be fine. The system runs, customers are served, everyone is happy.
But systems get stressed. Edge cases appear. Unexpected inputs arrive. And when they do, if nobody can read the logic, nobody can fix it. You’re standing over the oven, watching pizzas burn, with no idea how to adjust.
The solution isn’t to avoid AI tools. The solution is to never outsource your palate. Use the tools. Delegate the prep work. But always, always be the person who tastes the final dish before it goes out.
How to Taste the Pass
If you want to scale your work using AI without serving burnt food, the approach is straightforward.
Stop looking for complete automation. The goal is not to remove yourself from the process. The goal is to remove yourself from the mechanical parts so you can focus on the judgment parts. Delegate the repetitive work. Keep the quality control.
Use specialized tools for specialized tasks. Don’t ask one AI to do everything. Give each tool a specific station and a clear set of instructions. This is how you maintain quality at scale.
Invest in your palate. Whether you’re an engineer, a founder, or a business leader, your value is increasingly in your ability to recognize good output from bad. That means understanding the fundamentals, even if you’re not doing the manual work yourself.
The head chef who taught me to read the oven spent years on the line before he ever stepped up to the pass. He could taste a sauce and know exactly which ingredient was missing because he had made that sauce a thousand times himself. His expertise wasn’t replaced when he stopped cooking. It was concentrated into the place where it mattered most.
That’s the shift happening now. The question isn’t whether you’ll use AI tools. You will. The question is whether you’ll be the expert who knows what good looks like, or the operator who hopes the machine gets it right.
I’ll see you on the line.
Jacob