Guillermo Rauch Says AI Will Rewrite Most Human Made Software
We will all be learning in coming weeks & months, that most human-made software: ▫️ Was really slow ▫️ Had many security vulnerabilities ▫️ Was under-specified and incorrect AI will rewrite it all. Many times. It will make it fast, cost-efficient, and secure. It will think about failure modes we couldn't wrap our heads around. It will prove, formally verify, load test, harden. It will turn memory-unsafe foundations into robust, optimized ones. If you ask, it'll thoroughly document it, teach you about it, make diagrams, animations, and slide decks. In many ways, we are witnessing the true birth of software engineering.
Context
A widely shared take from the Vercel CEO frames AI as repeatedly rewriting software for speed, cost, and security, which shapes how builders talk about the next few months of tooling.
Guillermo Rauch argued that most human made software will be revealed as slow, insecure, and under specified, and that AI will rewrite it many times to make it fast, cheap, and secure. The post is commentary, not a product launch. It circulated alongside his Vercel Security Review results the same evening.
Deep dive
Guillermo Rauch (@rauchg) posted that people will learn that most human made software was slow, full of security holes, and under specified, and that AI will rewrite it repeatedly. He says models will prove, formally verify, load test, and harden code, and will document and teach the result when asked.
He calls this the true birth of software engineering. The post is a strategy and culture take from the CEO of Vercel, not a new benchmark or shipping product.
It landed the same evening as his Vercel Security Review 24 hour results, which gives the opinion a concrete product backdrop without turning the opinion post into a launch announcement.
Treat it as influential commentary that frames how one major platform CEO wants the industry to talk about AI assisted engineering.
What to expect next
Watch whether Vercel and peer platforms ship more rewrite and verify loops that match this rhetoric, and whether competing CEOs answer with their own framing. The useful test is shipping products and measured security or performance gains, not the virality of the take itself.