How to Write LLM Evals for Your AI Product
Most teams ship AI features on vibes. A lightweight eval framework — 10 golden inputs, a clear rubric, one scoring pass — changes that in an afternoon.
The unfiltered journey of building an AI-powered product — the bets, the rewrites, and what actually shipped.
All articles
Most teams ship AI features on vibes. A lightweight eval framework — 10 golden inputs, a clear rubric, one scoring pass — changes that in an afternoon.

Vibe coding lets you build software by describing what you want in plain language. Here is what it is, where it shines, where it bites, and how to do it well.

A user told us long sessions made Codepet lag and eat memory. We knew exactly what to fix, and the measurement proved us completely wrong. A short engineering story about resisting the obvious.

The honest story of building an AI-powered macOS app: the bets that paid off, the features we deleted, and what we'd tell anyone shipping with AI today.