We have entered a new era. The software engineer is no longer the one who builds. The software engineer is the one who decides what gets built, and how to direct the AI that builds it. Think of it like this: you are a naval commander. The crew is the AI. You are sitting behind the instruments, looking at the map, understanding the currents, the weather, the destination. You are not climbing the mast or hauling rope. You are deciding when to turn, when to accelerate, when to drop anchor. You are watching for what is wrong and what is right. You are the one who knows the difference.
What Changed
For thirty years, a great engineer was measured by depth. How deep could you go into the details? Could you optimize the algorithm? Could you trace through the stack? Could you hold the whole system in your head? The deeper you went, the more valuable you were.
That era is ending.
Today, a great engineer is measured by something else entirely: judgment. The ability to look at a large, complex problem and know which part to focus on. The ability to read code or a design and feel, in your bones, whether it is right or wrong. The ability to know when to intervene and when to let the executor run.
The details are not gone. They are just no longer the scarce skill.
The Three Dimensions of the Commander
A great commander operating an AI executor needs three things:
1. The Ability to Spot What Matters
You are staring at a thousand decisions. The AI can implement any of them. But which one actually moves the needle? Which one solves the real problem versus the symptom? Which one unblocks ten other things?
This is not a technical skill. It is pattern recognition. It is knowing your product, your users, your constraints, and your velocity deeply enough that when you see an option, you know instantly whether it is the right one. You know what is a priority and what is noise.
2. The Ability to Recognize Right and Wrong
The AI will show you code. It will show you designs. It will show you architecture. You need to know, quickly, whether it is right.
Right does not mean it compiles. Right means it is elegant. It is sound. It handles the edge cases. It is maintainable. It fits the style of the codebase. It does not over-engineer. It does not have hidden coupling. It does not create debt you will pay for later.
Wrong is often subtle. A solution that works but is inefficient. A pattern that makes sense today but will break at scale. An abstraction that is one level too deep. A naming choice that will confuse the next person. You need to feel these things.
3. The Ability to Know When to Intervene
The executor runs. Most of the time, you let it run. You do not interrupt every step. That is micromanagement and it defeats the purpose of having an executor.
But some moments demand intervention. The executor is heading down a path that will create a mess. You see it three steps ahead. You stop it now, at cost X, or you let it run and fix it later at cost 10X. The great commander knows the difference.
Intervention is also a communication. When you intervene, you are not just stopping something—you are teaching. You are showing the executor what right looks like. You are shaping its future work.
What This Means for Your Skill
If you spent the last decade optimizing for depth, you need to reorient.
You do not need to memorize APIs anymore. You do not need to know the exact syntax. You do not need to be able to hand-write a binary tree in an interview. These things are now table stakes that any executor can do.
What you need instead:
-
Read everything. Not deep reads into one system. Wide reads. You need to know how your product works, how other products in your space work, what the user needs, what the constraints are. You are building a mental map of the landscape.
-
Know what good looks like. Read great code. Read great designs. Read great systems. Build a sense for elegance and soundness. When you see something, you should know if it matches the standard or falls short.
-
Practice saying no. Or more precisely, practice saying “that is not the thing that matters.” Redirect constantly. This is a leadership skill, not a technical one.
-
Learn to move fast between contexts. You are no longer spending three months deep in one system. You are overseeing five systems, ten decisions, twenty trade-offs. Your brain needs to shift contexts constantly and still maintain judgment.
-
Build taste. Taste is the thing that cannot be executed. Taste is what the commander brings. It is the thing that makes a product feel right, even if you cannot articulate why. It is why people choose your product over the technically superior competitor. It is why users trust your judgment about what to build next.
The Uncomfortable Truth
Not everyone is cut out for this.
Some engineers want to go deeper, not wider. They want to spend a month understanding the garbage collector. They want to craft algorithms. They want to own a system so thoroughly that they are the only one who understands it.
That is still valuable. But it is no longer where the energy of the field is moving. The energy is moving toward judgment, toward taste, toward knowing what to build and why.
If you are thirty years old and you want to spend the rest of your career going deep, you can. But you will be swimming against the current. The industry is moving toward commanders and executors. The leverage is shifting away from depth and toward judgment.
The Role That Remains Scarce
I said earlier that as each layer of work gets automated, the bottleneck moves. The last thing that stays scarce is always the hardest to automate.
Right now, the scarce thing is decision-making. Judgment about what should exist.
How long until an AI is better at that than you?
I do not know. Maybe five years. Maybe twenty. But it is coming.
The AI will learn to propose features by reading your users, your metrics, your product roadmap, your competitive landscape. It will suggest to you that you should build X. You will read it and think, “I should have seen that.” And you will approve it.
What comes after that is just taste. Pure aesthetic judgment. And that might be the last thing that stays scarce, because taste is not optimization—it is creation. It is the ability to make something that is not just functional but beautiful. That is not just correct but elegant. That is not just useful but inevitable.
Maybe that is where humans stop competing on raw engineering and start competing on art.
Or maybe I am wrong. Maybe the commander role is enough for a lifetime. Maybe judgment is a skill that cannot be fully automated because it requires understanding humans, and humans are endlessly complex.
I am not sure.
What I am sure of is that if you are an engineer today and you are not noticing this shift—if you are still optimizing for depth and technical prowess and the ability to go deep into systems—you are preparing for the last era, not the next one.
Learn to be a commander. Learn to spot what matters. Learn to recognize right from wrong. Learn when to intervene.
The ship is getting faster. The executor is getting smarter. Your job is to steer it right.