AI writes correct Laravel now. Idiomatic code is the real test
Frontier models now pass Laravel's Boost benchmark. The harder test is whether the code is idiomatic, and most stacks can't measure that.
Tests that protect what matters: tenant isolation and reviewed, AI-assisted code.
Frontier models now pass Laravel's Boost benchmark. The harder test is whether the code is idiomatic, and most stacks can't measure that.
Copy-paste fixes, production hacks and "we'll refactor later" came first. AI just speeds slop up, and now agents write it too.
How we use AI coding tools every day, and the review rules that keep every change owned by a senior engineer who can explain it.
Row-level scoping or a PostgreSQL schema per tenant? How we choose, the code that enforces it and the tests that prove no tenant can see another's data.
Newsletter