90%
October 3, 2025
Three months ago I said that AI changes everything. I came to that after plenty of skepticism. There are still good reasons to doubt that AI will write all code, but my current reality is close.For the infrastructure component I started at my new company, I’m probably north of 90% AI-written code. I don’t want to convince you — just share what I learned. In parts, because I approached this project differently from my first experiments with AI-assisted coding.
When we hear about AI writing significant percentages of the code in a codebase right now, mostly that has been coming from people selling coding tools or people with something of a dog in the fight around this.
When we hear people sceptical about large language models writing code, very often this comes from a place of belief rather than significant experience.
When I first started using life language models as a software engineer, the better part of three years ago, I found them both astonishing and not particularly good. If I were inclined to criticise, I would talk about the loops they would get in where they would continue to make the same mistake over and over again. But what I saw, as others around me did, was something that showed genuine promise. Something that you needed to learn to work with and something that very quickly started to improve.
So I think the people best to listen to when it comes to this question of how well large language models can help develop software are the people who are doing it at scale – people who are very experienced software engineers who may well have started as being quite sceptical but who have seen the results of that. One such person is Armin Ronacher. Armin is the developer of a number of significant open source projects, including Flask. So someone who has earned the right for us to take what he has to say seriously.
When he says “AI is writing 90% of the code I was in charge of” Then this is something to pay attention to.
