A former software company founder visited a friend's 15-person tech team and witnessed a startling shift in development practices. Having recently shut down their own three-person shop, they observed developers treating Claude as the primary coding interface rather than traditional text editors and IDEs.

The author noted several concerning patterns: developers run multiple Claude sessions simultaneously without examining generated code, human code review has disappeared, and LLM-generated tests now flood repositories. Traditional software engineering principles around lines of code and abstractions appear irrelevant to this team.

Most troubling, deep problem understanding gets delegated to AI assistants. The author questions whether this represents a broader industry trend where programming transforms from a precise craft requiring deep knowledge into a probabilistic exercise of prompt engineering.

If these observations reflect wider adoption, the profession faces fundamental changes. Software development risks becoming a discipline where developers rarely engage directly with code logic, potentially redefining what it means to be a programmer in the AI era.