Lesson 2: The four things every good instruction has
Vague instructions produce vague recommendations you then have to fix. Almost every good instruction to a bee contains four things. Remember them as Goal · Scope · Limit · Why.
The pattern
- Goal — what outcome you want. Raise rates. Draft a reply. Dispatch cleaning.
- Scope — which rooms, dates, staff, or guests it applies to. Deluxe rooms, this weekend.
- Limit — the boundary you will not cross. Never above NPR 9,000. Keep 3 rooms sellable. Under a NPR 5,000 budget.
- Why — the reason behind it. Chhath rush. Slow midweek. A VIP is arriving.
The first three make the recommendation correct. The fourth — the Why — is the one people skip, and it is the most valuable, because the learning loop remembers your reasons and turns the repeated ones into confirmed policies (Lesson 6). A bee that knows why you did something can apply the same judgment next time without being asked.
Before and after
Vague: "raise prices"
The AI has to guess the rooms, the dates, how far to push, and why. You will spend the next three cards correcting it.
Clear: "For this weekend's Chhath rush, raise Deluxe rates but never above NPR 9,000, and keep at least 3 rooms sellable so we don't shut out walk-ins."
Goal (raise Deluxe rates), Scope (this weekend), Limit (≤ NPR 9,000; ≥ 3 rooms open), Why (Chhath demand, protect walk-ins). One card, approved.
You don't need all four every time
Short, well-scoped requests are fine — "flag any checkout-ready rooms for the 2 pm arrivals" is a good instruction. The pattern is a checklist for when a recommendation comes back wrong: nine times out of ten, a missing Limit or Why is the reason.