I used to think AI to be good at repetitive high level tasks, and I was right, until I wasn’t anymore. Over the past 3 months I tested LLMs on: 1) compiler and language design in TS 2) highly distributed systems in Elixir 3) kernel level programming in Rust
12
4
1
87
I was outperformed in any of the tasks by a wide margin the output is more comparable to a well oiled team of 10 over months than a solo over weeks

Nov 4, 2025 · 9:29 AM UTC

3
1
1
42
I started this being sceptical, I am now fully convinced code will quickly become a commodity, what database library or UI framework we use won’t matter anymore, the only differentiation is the programmer ability to understand design decisions deeply and skills in using models
4
2
1
70
The only moats in code will be 10x-ers that enable models to reason better, here I think Effect has a shot at solving the AI slop crisis
5
1
1
43
Replying to @MichaelArnaldi
Who do I pay to learn how to achieve such performance?
Replying to @MichaelArnaldi
Hello dear
Load more