"Which work is mine — and which can I hand off?"
Delegation is the first and most strategic of the four principles, because it is the only one that gets decided before AI is involved at all. It is the question you ask before you open a chat window, before you draft a brief, before any of the technique-level work begins: is this task one that genuinely benefits from AI involvement, or does it require something AI doesn't have?
That something is usually one of three things — judgement formed by years of experience, relationships built across years of trust, or authority that comes only from your role. AI carries none of these. What it carries instead is speed, breadth, and a tireless willingness to draft, restructure, and iterate. Good delegation is the act of matching one to the other. Bad delegation pulls in either direction.
Most professionals get this wrong in both directions, and they get it wrong consistently. Some delegate too little — treating AI as a party trick, opening it for novel questions and closing it again, never letting it carry real working weight. The cost is invisible: hours quietly spent on work that could have been done in minutes. Others delegate too much — handing over decisions that required their authority, their reputation, or their direct knowledge of the situation. The cost there is more visible, and more damaging when it surfaces.
The framing that helps most is this: delegation is not about maximising what AI does. It is about maximising the quality of your contribution by clearing the path for it. If AI can handle the research, the structuring, the first draft, the formatting, the supporting analysis — then your hours go to the parts that genuinely need you. The strategy. The relationship call. The judgement. The decision.
For senior professionals, this principle alone often produces more impact than any prompt-level technique. Not because the technique work is unimportant, but because most of the value comes from doing the right work in the first place. AI is most valuable to professionals who already know what their best contribution looks like — and who can recognise the path that protects it.
Treating AI as novelty rather than collaborator. Doing work by hand because that's how you've always done it, even when AI could compress hours into minutes. The cost is invisible — quiet, slow, but enormous over a year.
Handing over decisions that required your judgement, relationships, or authority. Treating AI as a substitute for senior thinking rather than a multiplier of it. The cost is visible — and damages your reputation when the work goes out under your name.