Using AI to write better code more slowly | Hacker News

Hacker News users said complex coding works better when AI is used across planning, implementation, review, and testing, not just in one prompt.

Commenters described combining models with different strengths: one writes code, another spots bugs, and later passes tighten architecture and edge cases.

The thread emphasized iterative, model-to-model feedback loops as a practical way to improve code quality and maintainability.

It also noted a limit of agentic coding: when developers must make many small design decisions, AI-driven workflows can slow down more than they speed up.