AI-assisted development shifts engineering focus to complex decision-making

The integration of artificial intelligence into software development is shifting the workload of engineers from writing syntax to managing complex product decisions. While AI models like Claude's Opus can rapidly generate functional code and handle 'happy path' scenarios, they struggle with edge cases and high-level architectural logic.

Developers are finding that AI can catch edge cases and even push back against incorrect instructions, but the technology cannot yet decide how a product should behave when unexpected user actions occur. This creates a new category of work termed 'the leftover hard parts,' where human engineers must determine product intent, user experience, and long-term technical promises—decisions that lack neat specifications and often require context from customer conversations or historical team decisions.