Agents Propose,
Humans Judge

The wager

Nursing solved AI delegation before AI existed

What the world calls "human in the loop," nursing calls supervision — and has case law about. What the world calls "AI safety levels," nursing calls scope of practice. What the world is groping toward with model cards and audit trails, Florence Nightingale was doing with a lamp and a ledger in 1855.

This is not a book about AI. There is no chapter on how the models work, and nothing in it expires with next quarter's release cycle. It is a book about the architecture of governed work — written by a nurse who has also been a physician, a hospital administrator, and an AI adoption consultant, and who has watched every wave of confident technology arrive at the bedside ungoverned.

The centerpiece

Chapter 13: The Five Rights of AI Delegation

Two scenes, eleven seconds apart in spirit: a night-shift nurse delegating a glucose check with an entire theory of governance compressed into a single handoff sentence — and a marketing manager typing "write the apology email and get it out before nine" to an AI that does exactly what it was asked. The difference between them is not carefulness. It is doctrine.

Every organization adopting AI is about to need people who can unbraid tasks from judgments, size circumstances, validate delegatees, give closed-loop direction, and supervise without rubber-stamping. There is exactly one profession that drills those five moves until they run in eleven seconds — and it is not computer science.