CAPPATERY counts down your daily bake before it goes stale, keeps the counter queue and POS moving, greets regulars with the order they always place, and works out tomorrow's bake quantities from what actually sold - not from what you baked last Tuesday.
Runs on the terminal and receipt printer already on your counter. No new hardware to buy on day one.
The morning rush board
One screen your counter staff can read at a glance mid-order.
Illustrative data on a demo counter. Levels drain, the queue moves, regulars surface as they reach the till.
Today's bake
Left / baked
Counter queue
Watching the till for a returning face
Counted continuously
Every tray, from first bake to close
Recognised at the till
Returning faces and their usual
Forecast nightly
Per item, for tomorrow morning
None of these are software problems on the surface. They are counting problems, memory problems and prediction problems - which is exactly what software is for.
Half the croissants go in the bin at close because no one tracked what was actually moving through the afternoon. The loss is real, it repeats daily, and it never shows up as a line item anywhere.
Wasted units are only ever discovered after they are unsellable.
Someone on their forty-first visit still gets asked what they would like. The order is identical every time. Asking costs fifteen seconds, and fifteen seconds repeated down a queue of eight is the whole rush.
Recognition lives in whichever member of staff happens to be on shift.
Quantities get set from memory and mood. Bake too much and it is waste, bake too little and the case is empty by ten with customers walking out. Nothing about last month informs tonight’s decision.
The same guess is made roughly three hundred and sixty times a year.
Track what is in the case, serve the queue quickly, project tomorrow from what really happened today, then catch whatever is left before it turns into waste. Each stage feeds the next.
Each baked line gets a quantity when it leaves the oven. From then on the count falls with every order taken at the till, so the number on screen is the number in the case - not a figure from the morning prep sheet.
Track / 11:20
Common orders sit on quick keys, payment splits without a detour, and when a returning customer reaches the till their usual is already staged. Staff confirm instead of interviewing.
Serve / counter 1
Overnight, each baked line is projected forward on its own curve using its own sales history, the day of week, and how it behaved on comparable days. You get a number per item, not a blanket batch size.
Forecast / tomorrow
As closing approaches, anything sitting well above its usual sell-through gets flagged with enough runway to do something about it - a bundle, a counter feature, a marked-down last hour.
End of day / 17:40
Plenty of software will take a payment for a croissant. Very little of it understands that the croissant expires tonight, that the person buying it comes in every weekday, or how many to have ready tomorrow.
Each of these runs on your own counter history. There is no shared model trained on somebody else's café deciding how much banana bread you sell on a Wednesday.
Standard inventory assumes goods keep. A bake tray does not. Every baked line carries a same-day life, a quantity that only falls, and an aging threshold - so the system can tell the difference between forty units in a stockroom and forty croissants that stop being sellable at six.
Low and aging thresholds are set per line, because a danish and a sourdough loaf do not age at the same rate.
See how this worksA returning customer is matched at the till, and their established order is staged for confirmation. It works from repeat-order patterns rather than a form somebody filled in, so it holds up when the person who normally recognises them is off that day.
Staff confirm or adjust. Nothing is charged from a prediction alone.
See how this worksEach line is forecast on its own history: how it sold on the same weekday, how it responded to weather and local calendar, how it trends month to month. Sourdough behaves nothing like a cinnamon roll and the numbers reflect that.
Forecast quality improves with history. Early weeks are treated as suggestions, and we say so on screen.
See how this worksSell-through is compared against the pace that line normally holds. When something falls behind with hours still on the clock, it gets surfaced with options - bundle it, feature it at the counter, or mark it down for the last hour.
The alternative is finding out at close, when the only option left is the bin.
See how this worksSelling both the same software and hiding the difference behind a pricing page does neither of them any favours. The product presents differently depending on which one you are.
Set up in an afternoon, run by the person who also makes the coffee.
The whole product is reachable from a single screen. Bake lines, quick keys, regulars and the nightly forecast - and nothing else demanding attention. No modules to license, no cost centres to configure, no approval chain for changing a price.
Central production, several counters, one standard held across all of them.
Each outlet keeps its own forecast, because a station branch and a residential branch do not sell the same things at the same hours. What is shared is the recipe of bake lines, the loyalty identity, and a rollup that lets you compare waste and sell-through side by side.
Trays, minutes at the till, units at risk before close. These are the things the product reports on from day one.
An honest note on results
We are not publishing waste-reduction percentages yet. Figures worth quoting need enough counters running for long enough to mean something, and when we have them they will name the conditions they came from.
Real time
Case counts move with the till instead of being reconstructed after close. The number you act on is the number in front of the customer.
Per visit
Matching happens at the till and is staged for one tap. Recognition stops depending on which member of staff is on shift.
Per item, per day
Every baked line gets its own projected quantity for the next trading day, on its own history rather than a shared batch rule.
Before close
Slow-moving lines are flagged against their own normal pace, with enough of the day left to act on it.
Two answers and the board fills in with a day that looks like yours.
Nothing to install and nothing to sign. Pick what your case leans on and roughly how busy the counter gets, and the rush board rebuilds around it - tray sizes, drain speed, queue depth and the kind of regular who would be recognised.
Today's bake
Left / baked
Counter queue
Watching the till for a returning face
Tray units to plan
72 a day
Across every baked line on this board
Queue depth at peak
4 deep
Roughly 160 seconds to clear
Forecast granularity
4 lines
Each one predicted on its own curve
This board runs on real numbers the moment yours are in it.
Sign up and we will map your bake lines, then send login details across.
