Definition
A style of software development where a programmer describes what they want in plain language and an AI agent writes, runs, and iterates on the code largely autonomously — the human steers by feel rather than writing every line. The UK's NCSC has published a risk framework specifically addressing the security governance of this practice.
Why it matters
Code produced without careful human review can contain security flaws, exposed secrets, or logic errors at scale; organisations need clear oversight policies to govern how much autonomy AI coding agents are granted based on how critical or public-facing the resulting software is.