Chapter 2
Why Vibe Coding Fails
The most instructive failure in the history of AI-assisted software development lasted six and a half minutes. In a controlled experiment designed to test whether multi-agent AI architectures could build a realistic software system, a pipeline of eleven specialized agents, each responsible for a specific stage of development from diagnosis through decomposition through implementation through review, was given the task of building a microservices backend for a vacation rental booking platform. The system had a budget of fifty dollars and access to the same language model that, in a different configuration, had scored a perfect twenty-eight out of twenty-eight on the same task. The pipeline spent its entire budget ...
Become an O’Reilly member and get unlimited access to this title plus top books and audiobooks from O’Reilly and nearly 200 top publishers, thousands of courses curated by job role, 150+ live events each month,
and much more.
Read now
Unlock full access