Fills a freed reformer down the waitlist by text, and handles the reverse case when a machine goes out of service and capacity drops.
An AI agent for reformer waitlists watches capacity in a Pilates studio and works it in both directions: when a client cancels it offers the freed machine down the waitlist, and when a machine goes out of service it identifies who has to be moved and moves them. The second half is what makes this different from a mat-based waitlist. A reformer is equipment, and equipment breaks — a snapped spring, a frayed strap, a carriage that needs servicing takes one machine out for three days. A ten-reformer studio becomes a nine-reformer studio, and every class that week is one place overbooked. Somebody has to work out who to move and where to.
Handles both a freed machine and a machine that disappears.
Offers a freed reformer to the waitlist by text within a minute of the cancellation
Holds the machine for whoever is being asked, so two clients cannot claim one carriage
Shortens the accept window as the class gets closer
Recalculates every affected class when a machine is taken out of service
Contacts the clients who have to be moved and offers them specific alternatives first
Restores the machine to the schedule when it comes back, without anyone remembering to
Most waitlist tooling assumes capacity is a number somebody typed in once. In a reformer studio it is a count of working machines, which changes without warning and affects every class until the part arrives. The manual version of handling that is a bad afternoon: opening each affected class, deciding who to bump, and making an awkward phone call, usually choosing the most recently booked client because that is the only defensible rule anyone has time to apply. Automating it does two things — it makes the choice consistent, and it turns the message from an apology into an offer, because the alternative slots go out in the same breath.
A three-step flow that treats capacity as a live number.
The agent tracks working reformers against booked places for every class, rather than assuming a fixed cap.
A cancellation triggers an offer down the waitlist with a held machine. A machine out of service triggers the reverse: identifying who must move.
Whichever direction it is, the client gets a specific offer — a place now, or two alternative classes — rather than a notification to act on later.
A realistic week where a reformer goes out of service on a Monday.
Scenario: a ten-reformer studio was handling equipment failures by opening every affected class in the booking system and phoning whoever seemed most movable. On Monday morning a spring housing on reformer seven fails and the part is three days out. The agent takes machine seven out of the count and recalculates: eleven classes between Monday and Thursday are now one place over. For each, it applies the studio's rule — most recently booked client moves — and messages those eleven clients with the reason and two alternative classes, before any of them have arrived. Seven accept an alternative in the first hour; three ask for a credit instead, which the agent applies; one asks to speak to somebody and goes to the manager. On Thursday afternoon the part is fitted, the machine goes back into the count, and the freed places reopen without anybody reopening them.
Studios whose capacity is equipment rather than floor space.
Equipment failure is routine, and the response is currently an afternoon of manual rescheduling.
Gives one consistent rule for who gets moved, instead of a decision made under time pressure.
Popular reformer classes have people waiting every week, and a freed machine is pure revenue.
Teach a class where the number of clients matches the number of working machines.
A client bumped from one room can often be offered the same class in another, if anybody checks.
Planned servicing has the same effect as a breakdown and can be handled the same way in advance.
Connects the equipment count to the class schedule.
Holds class capacity and bookings, and receives both the waitlist fills and the moves.
An alternative system of record for capacity and waitlist order.
Carries both messages — the offer of a freed machine and the notice that a client must move.
Applies the credit or refund where a moved client would rather not take an alternative class.
Tells the team a machine is out and how many clients were affected, without a group text thread.
Logs downtime per machine, which over a year tells you which reformers are costing you classes.
Where a changing machine count creates work.
Questions about automating capacity that physically changes.
An AI agent for reformer waitlists watches capacity in a Pilates studio and works it in both directions: when a client cancels it offers the freed machine down the waitlist, and when a machine goes out of service it identifies who has to be moved and offers them alternatives.
Most studios use most-recently-booked, because it is defensible and easy to explain. Some protect members over drop-ins, or protect anyone who has already been moved once that month. The important part is picking a rule and applying it consistently, which is exactly what does not happen when the choice is made in a hurry.
Somebody has to tell it — an instructor marking the machine down in your system or a message to the agent. It cannot detect a snapped spring. What it does is take that one input and work out every consequence across the week, which is the part that takes an afternoon by hand.
Offer the class first and the credit if they ask. A client who is moved to another time is still coming; a client who is credited has been given their money back and a free evening, and some of them do not rebook. Both should be available, in that order.
Yes, and it is the same logic at larger scale — every class in that room for the affected period, every client in them. This is the case where doing it by hand realistically does not happen at all, and clients find out on arrival.
A window you set, shortening as the class approaches — fifteen minutes when the cancellation is two hours out, five inside the last hour. A long hold is only generous if there is still time to ask somebody else.
It can show which classes are consistently at capacity with a waitlist, which is the argument for buying another reformer. That is a slower and more useful output than the day-to-day filling, and it comes free from tracking the count properly.
Fills a freed reformer down the waitlist by text, and handles the reverse case when a machine goes out of service and capacity drops.