‘Vibe Coding’ Signals the End of Software Programmers

Vibe coding...where to begin?
There's a new kind of coding I call "vibe coding", where you fully give in to the vibes, embrace exponentials, and forget that the code even exists. It's possible because the LLMs (e.g. Cursor Composer w Sonnet) are getting too good. Also I just talk to Composer with SuperWhisper…
— Andrej Karpathy (@karpathy) February 2, 2025
This is a great tweet by Andrej, but this tweet is not identifying a whimsical new workflow for engineering. It's a bunch of fun, flowery words to say: "I'm zoning out, directing an AI model, and doing a lot of copy and pasting because the system doesn't yet have full access to all the information and permissions it needs."
So-called "vibe coding" is not coding at all. It's basically just product management. Sure, for a few more months you may need to provide some technical assistance when the models get stuck, but this won't last. In short order, frontier models will be full-fledged software engineers, operating at the most elite level. Somehow all of this is being interpreted as a new way to code and not a mortal threat to this occupation. It is.
Vibe coding is a brief step on the way to being disintermediated. You are the middleman in this story, and you're being cut out. Software engineering is in the process of being fully automated.
Well, you can just oversee stuff and be more productive, right?
Perhaps. Certainly some of you will become rich off these amazing new powers, but this doesn't seem like the correct takeaway for the whole population of coders. It seems to me that this technology benefits two main parties: owners of capital and product visionaries. I get the impression that many software engineers believe they are going to just slide into product roles for some reason. You need to ask the question: are you any good at designing products people love? Why would this rare skillset just be assumed? These are totally separate careers with no established correlation. You might even be less likely to succeed at this than the general population.
I guess the other coping narrative is that your technical expertise will still be needed somehow. Like the product people won't know what to tell the AI to make, and they'll just end up with broken, non-scalable systems. This is an absurd notion in my opinion, and a further indication of how most of us don't have a truly internalized belief in AGI. What the labs are building is an AI engineer, not some sort of narrow code generator. This engineer will advise the product people and visionary founders just like a CTO and senior technical staff. They will be extremely sharp, extremely fast, and extremely easy to work with. They won't lose interest or use "mouse jigglers" to pretend to be working. These systems will make talented creatives extremely productive, and they will dominate the software industry for the foreseeable future as far as I can tell.
The idea that the AI will become the engineer—and not just the helper—is emotionally unpalatable but the evidence strongly points to it. If you believe in AGI, you already believe this.
And now for the general reader. Software engineers are just the first, but not the last, knowledge workers to be touched by synthetic intelligence. We will all have to grapple with these questions to some degree, wherever we stand in skill and wealth. What happens to the product visionaries when models are better at that too? It may take longer, but there's no fundamental reason why that won't happen as well. The whole game board is being turned upside down.
None of this is meant to dance on anyone's grave, far from it. It's meant as both a sanity check for you and for me. Software engineering is extremely difficult and important work, with a grand legacy of achievement; but that doesn't make you invulnerable to the tides of change. This type of automation has never occurred before and we all need to be on alert. Software engineering is a highly paid profession. If fully automated by models, that skillset could drop to being almost worthless overnight, and that's not hyperbole. If you and your family are dependent on this career, I highly recommend you consider changes to your lifestyle and develop a plan for transitioning to another field. I hope I'm wrong, and it just makes everyone more productive.