Coding with AI feels less like syntax and more like management

Writing code with AI feels less like operating a machine and more like managing a team.

Developer Allen Bargi published a note explaining why working with AI models requires a mindset shift. Traditional code is fully predictable: same instructions, same output. AI is different. The same request can produce a clever shortcut or miss an obvious point. Bargi argues that getting good results requires classic management habits: sharing context, setting boundaries, and steering intent. The post gained 239 points and 167 comments on Hacker News.

Why it matters: Software engineering is shifting from giving computers line-by-line instructions to guiding system behavior through conversation. Success increasingly depends on clear communication: explaining why the work matters, defining what a good outcome looks like, and applying judgment when the output strays.

Know this: AI has no lived experience, accountability, or human judgment. Reusable instructions and concrete examples help reduce misunderstandings, but human oversight remains the anchor.

The technology is brand new, but the leadership skills required to steer it are not.