Zalt Blog

Addy Osmani's Beyond Vibe Coding argues you should be the editor-in-chief of your AI-generated code, not a passenger. Here's an honest review: who it's really for, what it gets right, and where it just repeats itself.

The context your AI is missing is the knowledge nobody wrote down: the exceptions, the unwritten rules, the hard-won cautions your best people carry in their heads. Capture that tribal know-how and an ordinary setup starts producing work that feels expert. Skip it and the best model still answers like a first-week hire.

What's actually inside Gene Kim and Steve Yegge's Vibe Coding book? Four parts, a FAAFO framework, and refreshingly honest stories about AI agents deleting tests and nearly wiping out weeks of work. Here's the real breakdown.

Vibe coding vs traditional coding, settled honestly: one authors software by hand, the other directs an AI to do it. Neither wins outright. Here is when to use each, and whether vibe coding even counts as real coding.

Most work on attention focuses on accuracy. “Attention That Listens Efficiently” asks a different question: how do you make attention fast enough to matter in practice?

Working with autoregressive generation loops? "When Autoregressive Loops Stay Friendly" explores keeping them fast without making them painful to work on.

Struggle with how Pydantic stays so easy to use despite its depth? “The Facade That Makes Pydantic Feel Simple” breaks down the idea behind that simplicity.

Lazy Pipelines, Fast Backends digs into how to keep data pipelines easy to write while still hitting serious performance in the backend.