Claude Code creator says developers should guard code quality in the AI coding era

As AI changes how software is built, Boris Cherny, creator of Anthropic Claude Code, says developers should not focus on writing every line by hand, but on protecting code quality. He publicly replied to a developer's email asking how to handle "AI slop code," noting that AI is shifting engineers from writing code to reviewing AI-generated code.

The email described two approaches: one is to keep coding as before but use AI to speed up, with code still able to pass review; the other treats code as a black box and only checks the output. Cherny says the two do not have to be mutually exclusive.

He notes that prototypes and one-off code can be treated as black boxes, but production code should meet a higher standard than human-written code. Anthropic uses code checks, end-to-end tests, fuzz testing, automatic code review and refactoring to enforce this.

Cherny says the real job of developers is to guard code quality standards. This new role is sometimes called a "mini CEO," where developers act more like managers deciding what code enters the system.