A Hacker News user describes a dramatic shift from skepticism to amazement after using AI coding tools on legacy repositories. The experience revealed massive productivity gains for boilerplate code, build tools, and refactoring tasks that previously consumed hours of tedious work. Java developers report being able to run automated fixes like "add checkstyle, run mvn verify and repair" while taking coffee breaks instead of manually fixing linter warnings.

Initially cautious about AI-generated changes, the user found that constant diff checking became unnecessary as tools proved reliable for routine tasks. The workflow now involves feeding "analyze code" outputs back into the IDE, with developers only writing source code when specifications become too complex to express in plain English. This represents a fundamental shift in how software gets built.

The discussion raises a critical business question: when competitors cut development teams by 90% using AI automation, should companies follow suit or maintain teams and leverage the technology to build superior products? The author argues that keeping skilled developers while using AI as a force multiplier could enable companies to "dwarf" competitors with vastly better offerings rather than simply cutting costs.