←back
JP/EN

Laying the Groundwork

Preface

This post is neither about technology nor a lecture on careers. It is simply an essay gathering the feelings I've had lately while building software alongside AI—words I can't easily sort out.

If you are struggling with engineering alongside AI, or feel a frustration you can't shake, I hope reading this helps, even a little.

Introduction

Thanks to the remarkable progress of AI agents, AI is being put to use in more and more situations these days.

The same goes for system development, architecture, and design. With tools like Claude Code, Codex, and Figma Make, much of the work that used to require doing by hand can now be handed off to AI.

Just two or three years ago, generative AI lacked practicality, and the most we could do was use it deliberately for one-off tasks. Its growth since then has been so dramatic that some now say, "At this rate, won't humans stop being involved in development altogether?"

But is that really true?

Will humans stop doing engineering and coding?

It may well shrink dramatically, but I believe that "engineering will not disappear from human hands; rather, we are entering an era in which the importance and responsibility of engineers stand out more sharply than ever."

This doesn't depend on seniority. It applies to juniors and seniors alike.

"The Era of Writing Code Is Over" and "The Era of Understanding and Reviewing Code Is Ending"

On January 20, 2026, Ryan Dahl, the creator of Node.js, caused a stir in the engineering community by declaring that "the era of humans writing code is over."

This has been said a thousand times before, but allow me to add my own voice: the era of humans writing code is over. Disturbing for those of us who identify as SWEs, but no less true. That's not to say SWEs don't have work to do, but writing syntax directly is not it.

https://x.com/rough__sea/status/2013280952370573666


Then, on February 11, GitHub CEO Thomas Dohmke sent ripples through the community by saying:

"The concept of understanding and reviewing code is a dying paradigm."

https://x.com/ashtom/status/2021255786966708280


These remarks aren't simply about things getting easier. They foresee the development process itself shifting from being "code-centric" to "intent-centric."

Let's dig into what they said.

The second half of Ryan Dahl's post reads:

That's not to say SWEs don't have work to do, but writing syntax directly is not it.

This suggests that the value engineers create is changing. In any era, code that is simply copy-pasted, or found with a quick search and pasted in, has never had value in itself. The value of "engineering" lies in analyzing the problem in front of you and writing code as the solution.

In other words, my reading is that the simple, mechanical task of "coding" will fade out of our work, and everything else we do will become more important.

Next, let's look at the first half of Thomas Dohmke's post.

In the last three months, the fundamental role of the software developer has been refactored. The incredible improvements from Anthropic, Google, and OpenAI on their latest models made coding agents so good, in many situations it’s easier now to prompt than to write code yourself. The terminal has become the new center of gravity on our computers again. The best engineers can run a dozen agents at once.

Yet, we still depend on a software development lifecycle that makes code in files and folders the central artifact, in repositories and in pull requests. The concept of understanding and reviewing code is a dying paradigm. It’s going to be replaced by a workflow that starts with intent and ends with outcomes expressed in natural language, product and business metrics, as well as assertions to validate correctness.

The rest of the post shifts to introducing his newly launched company, so you could also read it as a kind of position talk—speaking from where he stands.

What deserves attention is the line "The best engineers can run a dozen agents at once." Today, a wide variety of AI agents are offered not only in enterprise products but also in OSS, SaaS, and beyond.

It's becoming clear that, going forward, AI agents will be used as a matter of course, just like IDEs and extensions.

Accept the Reality You Face, and Think About What Comes Next

When I talk with others about combining AI and development, I get all kinds of reactions.

"AI is weak at such-and-such, so it's useless." "This AI agent's quality isn't good enough, so I can't use it." "Developing with AI is boring in the first place."

"If we leave everything to AI agents, do we even need readable code?" "Even what it can't do now, it'll evolve to do eventually." "People who say AI is useless for development just aren't using it well." "To make good use of AI, you have to be tolerant of it."

Each of these opinions is logic built up from that person's position, the projects they work on, and the people they spend time with, and every one of them struck me as a convincing argument.

All sorts of claims are being made from all sorts of standpoints, and I don't think any of them is simply wrong. If there's one thing I can say, it's this: "Technology evolves every day, and the world keeps changing."

Whether you adapt to the big change we're facing right now, or keep delivering value that holds steady no matter the situation, is up to you.

Eat the Poison, the Medicine, and the AI Too—and Firm Up Your Foundation

There is a page in the manga Baki with this line:

Preservatives... colorings... additives... all kinds of chemicals. There's no way they're good for the body. And yet, eating only what is healthy is hardly wholesome either. Eat the poison. Eat the nourishment. What matters most in eating is the capacity to find both delicious and turn them into flesh and blood.

I love this line, and whenever I'm unsure whether I should do something, at work or in my private life, I think back to it!

Just because AI is spreading doesn't mean I think devouring documentation and technical books is a waste. Nor do I think it's a waste to keep following AI technologies, papers, and new approaches.

Even if AI turns out to be slash-and-burn farming for engineering, why not devour even that, firm up your foundation, and bring out strengths that are yours alone?

The World Is Full of Position Talk—So Ride Along with the Talk That Sits Well with You

It's an extreme view, but I think every comment in the world is position talk. People are moved by how much the weight and background of words stir them, and by whether those words convince them.

Even a comment that objectively seems "pretty extreme" is often compelling and full of appeal to a certain group of people. (I'm aware that the words I'm drawn to probably sound extreme to someone, too.)

We live in a world where anyone can broadcast, and there is an enormous amount of information. What people say varies so much that it's often hard to know what to believe and which way to go. To put it bluntly: if you can't judge anyway, just ride along with the narratives that match your own values.

I believe a series of such choices is what creates value that is yours alone, and from here I'll keep moving forward, firming up the ground beneath my feet as I go.


Thank you for reading to the end.