A CTO who thinks software engineers have 'several years' lef

Gil Feig employs software engineers. He works with them every day. He also believes his profession is disappearing.

“I’ve seen such an inflection point in the growth of sort of intelligence of these agents that I’m not here to say it’s not going to happen anymore.”

This isn’t a headline from a research paper. It’s a founder who has shipping agents, talking about something he sees in his own codebase. AI is writing code. It’s writing good code. It’s writing code better than the median engineer he knows.

So I asked him the question everyone asks: How much time do we have?

“I think there’s still several years, right? There’s still many, many years. Because even if AI can write all the code, if you can choose between having software engineers backing that and supporting the AI and doing it and not having software engineers, why would you not opt for software engineers doing that right now? But I think over time, that argument gets weaker.”

The Baltimore bridge metaphor

Here’s where the conversation got tense. Gil offered a metaphor that he couldn’t unhear, and neither could I.

“I always think of that ship that hit that bridge in Baltimore and how for like 10 minutes you could see it was going to hit the bridge. Everyone knows like this is going to hit, but there is nothing you can do. This thing is just going, going, going. And it’s like a large keep a medal with so much momentum in a specific direction. That’s how enterprise is right now.”

The metaphor isn’t about the bridge. It’s about the moment. The ship’s engines are set. The momentum is locked in. You can see the collision point. But nothing stops it. Enterprise momentum doesn’t pivot on a dime. It hits obstacles, then adjusts.

That’s how AI and employment work. The trajectory is visible. The momentum is real. But institutions move slowly. They’ll change, “but they might have to hit the bridge first.” Something will break. Then change happens.

For software engineers, that bridge is the moment when AI-generated code becomes standard enough that companies can’t justify hiring humans at the same salary. That’s not tomorrow. It’s not even next year. But it’s visible from here.

What replaces the engineer

Gil doesn’t think jobs disappear. He thinks they transform.

“What’s really gonna emerge is a new type of role here that a lot of different professions are gonna move into, which is some form of operator of the AI.”

This is the thing people don’t want to hear. It’s not “everyone will be fine.” It’s “everything will change and you’ll need to learn new things.” Which is true, and also unsettling.

An “operator of the AI” is someone who writes prompts, structures workflows, handles exceptions, manages the AI when it goes weird. They’re not an engineer anymore. They’re not a manager. They’re something new. The salary might be the same, or it might be lower. The work is different. The bar for entry is different.

Some people will adapt. Some won’t. That’s not a prediction about intelligence. It’s a prediction about change management.

The economic anxiety underneath

I pushed back on something Gil said. “What can one do to be prepared?” It’s advice I’ve heard before. Upskill. Learn AI. Stay relevant. It all assumes you have agency in the transition.

“The things I’ve read are like, make all your money now because you’ll never have a chance to make money again, which is like so scary. I think a lot of this is a little bit crazy.”

That’s the loudest narrative right now. The doomsday version. Make your fortune before the robots take over. It’s not calming. It’s not practical. It’s fear.

But Gil acknowledged something harder: there’s real uncertainty. No one knows how this resolves. Governments might intervene. They might not. Companies might decide AI-generated code is good enough. Or they might decide you still need humans. The economics could shift in ways no one predicts.

“I think wherever you are in the political spectrum, you’re gonna see that jobs are gonna be hit. And so there’s gonna need to be government intervention in some way here. No one knows exactly, but I think everyone is pretty fearful of a post-AGI world.”

The question no one can answer

I asked Gil what he thought humans could do better than AI. What would remain?

“What is AI not going to be better than us at? And the answer is like probably nothing.”

He didn’t say that confidently. He said it like someone running the logic and reaching an uncomfortable conclusion. If AI keeps improving at the rate it has been, if the scaling laws hold, if hardware keeps accelerating, then the list of things humans are uniquely better at gets shorter. Maybe it hits zero.

But maybe not. Maybe creativity stays human. Maybe taste stays human. Maybe there’s something about embodiment or lived experience that keeps humans valuable. Or maybe those get solved too.

“I don’t know. Bring back the idea of human creativity. Who knows? Who knows how good AI would get at that?”

What this means for your career

If you’re a software engineer reading this, you have “several years.” That’s not nothing. It’s enough time to:

Get very good at prompt engineering and AI workflows
Learn how to evaluate and fix AI-generated code
Understand where AI still struggles (hardest problems, novel approaches)
Build a reputation in areas that require taste and judgment, not just execution

You have time, but not infinite time. And the transition is already happening. The engineers getting displaced first aren’t the ones getting better at AI. They’re the ones waiting for it to stop being true.

Gil’s honesty matters here. He’s not selling optimism. He’s not selling dread either. He’s watching it happen and describing what he sees. An inflection point. A momentum that’s hard to stop. Years before it lands. An uncertain landing.

“I think over time, that argument gets weaker. Software engineers just become really expensive. And at some point, it’s like, yeah.”

“Yeah” is the sound of acceptance without comfort. The bridge is coming. You can see it from here.

FAQ

Is Gil saying software engineers will disappear? He’s saying the profession will transform. Code-writing engineers might become AI operators. Or the role might become something else. But the raw need for people who know how to write code? That probably goes down.

How many years is “several years”? Gil said “many, many years” but also acknowledged he might be wrong. Probably between 5-15 years before AI is so good that hiring humans for code-writing is hard to justify. That’s his estimate, not gospel.

Should I quit being an engineer? No. Use the years you have to get good at the new skillsets (AI direction, evaluation, edge cases). The transition is slow enough that you can prepare. Waiting does nothing.

Will there be new jobs created? Probably. But they’ll be different. The jobs we have now might not exist. New jobs might be better or worse. There’s no guarantee they’ll be as lucrative or plentiful.

What’s the “Baltimore bridge” moment for software engineers? It’s the moment when AI is good enough and cheap enough that hiring humans becomes obviously wasteful. That moment might come fast once it comes. But we’re not there yet.

Can government slow this down? Probably not for long. But it could cushion the transition (retraining, UBI, whatever). Gil thinks it’s inevitable but also thinks institutions are slow. So there’s time.

What should I learn to stay relevant? How to work with AI as a tool. How to evaluate its output. How to handle edge cases it misses. How to think about problems AI struggles with. These are skills that will matter for the next phase.

Is the “operator of the AI” job as good as being an engineer? Unknown. It might be better (higher leverage, clearer impact). It might be worse (lower pay, less autonomy). The work will definitely be different.

Are there jobs AI won’t replace? Probably. Anything that requires embodied presence, judgment calls, taste, or novel creative work. But Gil was skeptical that these jobs will scale to replace the number of engineering jobs that disappear.