A thoughtful review of AI-generated frontend code in a real product: strong in isolated spots, but increasingly inconsistent at the system level. It also makes the case for using AI as a candidate generator and validator, not as the reviewer with the final opinion.
PHP
Reading code with AI, not generating it
A good piece on using AI to understand a large existing codebase instead of generating more code. It shares practical onboarding tactics and prompts that help map domains, dependencies, and architecture faster. Read more
