Another brick in the wall.

That was my first thought reading Paul Ford’s recent guest essay in The New York Times, “The A.I. Disruption We’ve Been Waiting for Has Arrived.” Not because he said anything outrageous. Quite the opposite. Because he described something that felt calm, practical and very real. And when calm, practical people start describing economic earthquakes, you pay attention.

Ford talks about “vibe coding,” a term coined by Andrej Karpathy. He describes riding the subway home, logging into Claude Code on his phone, uploading some files and typing prompts like, “Load this into a database and make it searchable with a web interface.” By the time his train crosses the Manhattan Bridge and pulls into Brooklyn, the app is largely built. A website. A tool. A small game. Something that might have sat in a folder for years is suddenly live.

Then comes the line that should have landed like a punch to the gut. He says as he watches his AI agent compile, he can’t help but see the faces of former employees. Designers. JavaScript developers. Engineers he once hired and billed out at serious money.

That’s when I had my Sixth Sense moment.

If you know, you know. But for those who don’t, in the movie The Sixth Sense, Haley Joel Osment’s character sees people who look normal at first. They talk. They move. They exist in the room. The twist is they aren’t alive.

When Ford says he sees those developers’ faces, I worry we are looking at the same kind of moment. Not because developers are suddenly irrelevant. But because if this keeps improving at the pace we have seen since November, some roles may not survive in their current form.

Ford isn’t speculating. He gives numbers. He rebuilt his own messy website and realized he would have paid $25,000 for someone else to do it not long ago. He cleaned up a complex data set for a friend and thought about how he might have billed $350,000 for that kind of project in 2021. That price implied a product manager, a designer, multiple engineers and months of work.

Now he pays $200 a month for Claude.

That is not incremental efficiency. That is cost structure collapse.

He describes November as a turning point. Before that, AI coding tools were useful but clumsy. After that, they could run for an hour and produce credible applications. Not perfect. Not handcrafted. But functional, on time and shockingly cheap. He talks about software stocks dropping, markets repricing, legal software firms taking hits because automation tools can now do portions of their work. He reminds us that “real artists ship,” and suggests that AI might be erasing the friction that kept many products from ever shipping at all.

Here’s the thing. Markets may overreact, but they rarely hallucinate for long. When you can compress $350,000 of labor into a $200 subscription, entire business models start to wobble.

This isn’t about whether the code is elegant. It’s about whether it works well enough. Most business software is not art. It is plumbing. If AI can build functional plumbing fast, leaders will not wait for marble.

I have been writing for a while now that AI is not going to replace you. Someone using AI better than you will. That line applies nowhere more directly than here.

Developers are not suddenly obsolete. But static developers are vulnerable. If your value proposition is that you can hand-code what a capable AI agent can now assemble in minutes, you have a problem. If your value proposition shifts to architecting, orchestrating, validating and securing AI-generated systems, you have a future.

This is where DigitalCxO leaders need to lean in.

If you are running digital transformation, product, engineering or innovation teams, vibe coding is not a curiosity. It is a signal. Your cost models are going to change. Your time-to-market expectations are going to compress. The classic build-versus-buy debate is about to get rewritten because “build” may not mean hiring five more engineers. It may mean empowering two strong operators with the right AI stack.

The org chart shifts too. Skill evaluation moves away from pure syntax mastery and toward problem framing, system thinking, and AI orchestration. The developer who can define the right prompts, validate outputs, understand architecture and anticipate failure modes becomes more valuable than the one who can manually write every function from scratch.

At the same time, let’s not kid ourselves. There is a psychological cost here. Software development has been a craft. A profession with identity. Pride in shipping something handcrafted. Ford admits that watching this happen is not altogether pleasant. He feels less valuable than he used to be. That stings.

And yet he is excited.

That tension is real. Many of the people I speak with feel it. There is fear. There is denial. There is bravado. But underneath it all, there is recognition. Since late 2025, coding agents have taken a visible leap. They are not perfect. They generate insecure code. They hallucinate. They burn data center water and electricity at alarming rates. But they are better. And improving.

There is simply too much signal now to dismiss this as hype.

You see it in enterprise software, adding AI build features. You see it in consulting firms quietly experimenting with smaller teams delivering larger scopes. You see it in early adopters building internal tools in days that once took quarters. You see it in markets reacting violently to perceived labor compression.

We have entered a new stage of AI coding capability. Not the end state. Not artificial general intelligence. But a meaningful phase shift.

The mistake would be to frame this as developers versus machines. That is lazy thinking. The real shift is between developers with AI versus developers without it. Organizations redesigned for AI leverage versus those clinging to old staffing assumptions. Leaders who reimagine workflows versus leaders who hope this wave passes.

It won’t pass.

Ford ends his essay saying he may be less valuable than before, but if this technology keeps improving, people who struggle with bad software could finally get the tools they deserve. That might be a good trade.

That is the leadership question.

Are you going to protect the old cost structures, or are you going to unlock new capability? Are you going to treat AI coding as a threat to contain, or as leverage to amplify your best people?

Because if you ignore this moment, if you dismiss it as noise, you may find yourself in your own Sixth Sense scene. Looking around at roles, processes and org models that feel familiar, comfortable and alive.

Until one day you realize they are not.

Developers who adapt and adopt will thrive. Those who refuse may fade. And the leaders who pretend this is still 2021 pricing and 2021 velocity are going to wake up to a very uncomfortable reality.

If Ford keeps riding that subway and watching his code compile, he may keep seeing faces.

The question is whether those faces belong to the living.

Or whether, one day, he looks up and says, very quietly, I see dead developers.