An LLM can write a rate limiter in seconds. It cannot decide how rate limiting should work in your system.
As models get faster, implementation gets cheaper. Judgement and the expertise behind it becomes the bottleneck.
hbish.com/ai-makes-deep-expe…
1
92
The "near miss" is strangely addictive. 🎰
We talk about AI as a tool for efficiency, but the real hook is something else entirely. It’s that feeling of being "almost there."
But that gap isn't a bug; it's the feature.
hbish.com/on-being-almost-th…
35
2025, was sitting in my draft. Decided to just post it.
hbish.com/2025/reflecting-on…
43
First blog post in yoinks!
My learning journey with the workman keyboard layout and thoughts after 5 years of daily use.
hbish.com/2026/workman/
61
