In the AI Era, What Should Developers Own?#

2026-03-18

As AI has advanced rapidly, my state of mind has also swung quite a bit over the past few years.
Before AI seriously entered the way we work, I was a very confident developer. I liked development itself, and I enjoyed learning new things. I saw myself less as a specialist and more as a generalist and full-stack engineer, and I believed my strengths were in structuring ideas logically and communicating clearly.

Then AI-based tools truly arrived. In particular, once I started using tools like Cursor, my productivity increased noticeably, and for a while my confidence actually grew even more. It felt as if the world was moving in a direction that favored people like me. The barrier to software development seemed to be falling, and it felt likely that there would be many more one-person startups. I was excited by the idea that individuals and very small teams could now do genuinely big things.

At that stage, I felt more excitement than worry. Even at work, I often thought that AI tools could help not only developers but non-developers do their jobs better, and I talked about that idea often. Whenever a new AI tool appeared, I was usually one of the first to pay attention. I openly believed that real working methods were about to change.

The moment that excitement started to shake#

Then, at some point, my feelings started to change.
I watched an operations colleague use Claude Code to build a surprisingly structured environment for their own work.

In that environment, a meaningful part of the workflow had already been automated: data collection and analysis, report writing and distribution, task management, ad operations, and performance analysis. What struck me even more was that this person did not stop because of technical barriers. They asked AI what they needed to ask, got guidance through blocked points, and kept refining their workflow.

The moment that hit me hardest was when I saw that person update data and handle work directly inside that AI environment instead of using the admin page I had spent time building. Intellectually, I had already believed that this kind of era could arrive. But seeing it happen in front of me triggered a very different emotional response.

That was the first moment when the idea really landed: the moat of “being the person who knows the technology well” might weaken much faster than I had expected.

My anxiety was not just one thing#

For a while after that, I felt somewhat deflated.
When I looked closely at that emotion, I realized it was made up of two different things.

• One was the sense that the world really is changing.
• The other was the feeling that my own relative scarcity was weakening in the face of that change.

The former is closer to reality recognition, while the latter is closer to an identity tremor.
The problem is that when the two get mixed together, your judgment gets blurrier and the emotion gets heavier.

There was a time when simply knowing programming syntax well, or understanding frameworks and cloud resources in detail, felt like a major strength in itself. Of course, that knowledge is not completely meaningless even now. But compared with the past, far more people can produce real results through AI without knowing those details deeply.

The ability to make things still matters. But it seems harder and harder to expect long-term security from that alone. Even structuring workflows, improving them through trial and error, and turning them into repeatable systems no longer feels like something reserved only for engineers.

So what remains?#

At that point, I felt I needed to change the question.

My old questions were closer to these:

• Will developers still remain important?
• Can technical skill still be a moat, even in the age of AI?
• Will people who design systems be in a stronger position?

But the question that feels more important to me now is this:

• At what point in the value created with AI does someone actually hold ownership?

Once I shifted to that question, my thinking started to change.
What mattered was no longer just “What can I build?” but “Can I directly capture a piece of the value that gets created?”

Why ownership matters more than the ability to produce#

AI lowers the cost of production.
It lets us create things with far less time and far fewer people than before.

But as the cost of production falls, the act of producing itself also becomes less scarce.
Features are copied faster, and visible outputs become easier to imitate.

At that point, the things that become relatively more important usually look more like this:

• The ability to decide which problem to solve
• The ability to understand the real context of customers or users
• Trust that makes people come back repeatedly
• The ability to create distribution and adoption
• Structures that accumulate data and relationships
• Structures that tie rewards directly to the value being created

In other words, what may matter more going forward is not “I can build well,” but “Does what I built actually get used, earn trust, accumulate relationships and data, and ultimately lead to money?”

My thinking eventually moved toward business#

As I kept thinking about this, my attention naturally began to move toward business.
Whether that means starting a one-person company, or taking on a role closer to a business partner than “just a developer” inside a company, what feels important is moving toward a position closer to being an owner.

By owner here, I do not necessarily mean only someone who founded a company.
The kind of ownership I have in mind is a bit broader.

• A position where I can directly capture part of the value that gets produced
• A position where I have influence over what gets built and which problems get solved
• A position where I share responsibility for outcomes, but also receive rewards tied to those outcomes

From that perspective, it starts to feel insufficient to remain only “a developer who uses AI well.” It is certainly important to have the confidence that I can build more with AI. But at the same time, it is becoming increasingly obvious that I am not the only one who can do that.

So I want to change the question like this:

What should I build? to What should I own?

The anxiety of an era where copying is easy#

There is another concern attached to this.
If AI gives me confidence that I can build almost anything quickly, it also creates the fear that other people can catch up at a similar speed.

At that point, it can feel like everything eventually becomes a marketing battle, and beyond that, a capital battle. For an individual, that can look like a very unfavorable game. That worry is entirely realistic.

Still, over time, I have leaned more toward the view that the real question is not how to prevent copying completely.
What matters more is building a structure that is hard to take away even if it gets copied.

That might mean things like:

• Solving a very specific customer problem
• Going deep into the real workflows of a particular company or industry
• Building trust with customers
• Creating a structure where repeated use and data accumulation happen
• Providing not just software, but also ways of operating and interpreting it

In the end, the moat seems more likely to come from relationships, context, and accumulation than from the feature itself.

The direction I want to hold onto now#

At this point, the direction I want to hold onto is actually quite simple.

First, I need to belong to the group that uses AI the best.
Since change cannot be stopped, it feels more important to stand on the side that handles these tools most skillfully, rather than keeping them at a distance.

Second, I need to see technology not as the end goal, but as a means of production.
It matters that I can build quickly, but I also need to check more often whether what I build turns into an asset.

Third, I should move as close as possible to the point where value is captured.
Inside a company, that means moving toward problems tied directly to outcomes. Outside a company, it means running small experiments that still carry ownership in my own name.

Fourth, I do not need to abandon the identity of being an engineer, but I also do not need to remain trapped inside it.
It may be a stronger path to keep my engineering instinct while adding business sense, operational sense, and customer sense on top of it.

Closing thoughts#

I still genuinely like development.
I still enjoy building things, shaping structure, and simplifying complex problems. So this is not an essay that wants to conclude, “Developers are over now.”

In fact, it is closer to the opposite.
The more the ability to make things spreads widely in the AI era, the more important it becomes to think about what you actually own on top of that ability.

The next stage I need to prepare for is probably somewhere in that direction.
Not stopping at being someone who uses AI well, but gradually becoming someone who directly owns part of the value created with AI. And eventually, standing not merely as someone who uses the tools of production, but as someone who holds part of the value flow itself.

I have not found the full answer yet.
But at least the question has become clearer.

What should developers learn more of in the AI era?
That is an important question too.

But for me, the more important question right now is this:

In the AI era, what should I own?