An appealing proposal meets a full bay
A repair-workshop owner needs reliable scheduling for paying customers. A broad AI assistant looks attractive: more features seem to promise more value. All figures in this hypothetical comparison are assumptions; neither service has a setup charge. The assistant uses completed jobs to propose Tuesday at 10:00 because similar repairs often started then. Its historical dataset contains no current bay status. Yet the live calendar marks the required bay occupied. Past patterns do not establish present availability.
What each option costs to operate
For 120 bookings per week, the broad assistant costs $60 weekly and needs five minutes of paid calendar checking per booking. At $30 an hour, 120 × 5 ÷ 60 = 10 review hours; $60 + 10 × $30 = $360 per week. A narrow service checks the live calendar and blocks occupied bays. Its fee is $180 weekly, with one minute of staff review per booking: 120 × 1 ÷ 60 = 2 hours; $180 + 2 × $30 = $240 per week. Both fit the owner's $400 weekly scheduling budget. Choose the narrow service if bookings pass the calendar check and testing supports these review times.
Test the occupied bay
Inspect a request with requested_slot = Tuesday 10:00 and the matching bay's current calendar. Run it through each option and time the staff's work. The broad workflow should catch the conflict during review; the narrow service should block it before confirmation. Staff then offer an available slot. If either confirms the occupied slot or uses stale availability, retain manual confirmation and repair the check. Think of each confirmed booking as a contract for a bay.
A quieter diary reverses the choice
Now suppose bookings fall to 30 weekly. Broad review takes 2.5 hours, so $60 + 2.5 × $30 = $135 per week. Narrow review takes 0.5 hours: $180 + 0.5 × $30 = $195 per week. With reliable manual checks, choose the broad option. Costs meet at 60 bookings weekly; higher volumes favor the narrow service, lower volumes favor the broad assistant.