Theory
The traditional model had engineers as primary implementers: they translated specs into code, wrote tests, and owned the mechanics of delivery. AI coding tools are shifting that.
The vibe-coding trap
"Vibe coding", describing what you want in plain language and letting the AI fill in the rest, works for demos and prototypes but not for production. The AI guesses; when you build on guesses, things break in surprising ways. The move to AI-Native Engineering is to treat AI as a partner you direct, not a magic box.
The new role
Engineers become orchestrators: they define intent (what we want and why), set constraints (architecture, standards, non-goals), and own verification (tests, reviews, quality gates). The agent handles first-pass implementation; the engineer reviews, corrects, and steers. Ownership of code, especially for new or ambiguous problems, still rests with engineers; agents take on the mechanical, multi-step work.
Practice
Run this in a repository you already know, not a toy project. The point is to feel where the practice helps and where it gets in the way on code that has history.
List three tasks you currently do that you would delegate to an AI agent first, and three you would always review or own yourself. Compare with a colleague if possible.