Walk into a workshop at 9am and you can usually spot the day's economics in thirty seconds: two bays empty while a queue of vehicles waits for the one lift that handles SUVs, a technician idle because his next job's parts are on a truck, and a service advisor promising "tomorrow, definitely" with no idea whether tomorrow has room. The bays are the bottleneck asset — the thing the whole business exists to keep busy — and they are managed with less rigour than the stationery cupboard.
A bay is a bookable resource
The fix is a reframing: a bay is a facility with a type (mechanical, denting, washing, the SUV-capable lift), a calendar, and a capacity — exactly like a technician is a person with skills, a shift and a load. Model both as bookable resources and the workshop's day becomes a scheduling problem with known arithmetic: a job needs a bay of type X and a technician with skill Y for duration Z; the engine finds candidates, checks overlapping bookings, and books both.
What the arithmetic buys you
- Promises with substance. The advisor quoting a delivery time is reading real capacity, not optimism. Slot-based booking means the customer chose a window the workshop can staff.
- Utilisation you can see. Bay occupancy by type and hour is a report. The case for the second alignment bay arrives as data, not as a foreman's frustration.
- Honest interactions between waits. A job waiting on parts visibly holds or releases its bay — the difference between one stranded vehicle and a stranded afternoon.
- A day that re-plans. When the 11am runs long, drag-and-drop on the Gantt re-flows the afternoon — and everyone's screen agrees, because there is one schedule.
The same discipline, three surfaces
The insight generalises: the demo vehicle on a test drive, the pickup driver, the delivery slot at handover — all bookable resources with the same capacity arithmetic. A dealership that learns the discipline once applies it everywhere customers are promised a time. The whiteboard survives as decoration.
Deterministic on purpose
None of this is AI, and that is the point. Scheduling promises must be explainable — "why is my car not ready" deserves a better answer than a model's vibes. The engine is inspectable arithmetic: skills, types, capacity, overlaps. Where agents help is around the edges — spotting the chronic Tuesday overload, drafting the re-plan when three no-shows open the morning — with the scheduler still deciding.
In the product: workshop management and appointment scheduling & capacity planning.