💭 in your head · nothing to type
Think of a task you recently handed to AI — say, "write me an Instagram post."
Don't dissect the request. Picture it as the turning of seasons. Winter is a command with no context: a bare branch, no leaf, no fruit. Spring is the first hint of a topic. Summer is full context — audience, goal, tone, length. Autumn is iteration, refinement.
Run it in your head: which season was your request in? What hadn't bloomed in it yet?
💭 in your head · nothing to type
Now shift the frame: not seasons — the changing of the guard. One sentry stands exactly as long as ordered and does precisely what's in the manual. The other was told why they're posted here — and sees the perimeter differently.
Hold this in your head: which of your tasks is AI currently "standing guard" on without knowing why? Where are you issuing an order instead of passing along the meaning?
The more "why" in a prompt — the less guessing in the answer.
This isn't "AI doing it instead of you" — these are four ways of co-working. They're different facets of how you and the agent think and build together. There's no single right one — each fits a different task.
Before: Write a function
Now: I need a short, friendly Instagram post about our new service, with a call to message us
The agent isn't a robot you give commands to. It's a teammate you brief on the task.
<aside> 🧭
Interactive block — on the site: open →
</aside>
The first answer is a draft, not a verdict. A bad prompt is data, not failure. Iterate.