Cover decay
Compares cover today with cover in ninety days, counting only lots that will still be in date. The product's central fact and its first check.
Flags when the fall exceeds the threshold your household set.
Auto Prepped · Auto Prepped
Photograph what is in the cupboard. Auto Prepped reads the dates, works out how many days you could eat from it, and tells you what to use next so the number holds.
Capture
Four steps, in this order, for every item.
One shot of the barcode, one of the printed date. Six identical tins are one photograph and a count.
Identity bounds which dates are plausible, and the date confirms the identity. They resolve together.
A medium-confidence read shows both images and asks one question. A low one waits in a queue. Nothing is guessed.
The lot joins the pantry with its date, its precision and where the date came from, and the number updates.
Pantry checks
Every check states what it watched and what it can and cannot conclude. Results close themselves when the condition clears.
Compares cover today with cover in ninety days, counting only lots that will still be in date. The product's central fact and its first check.
Flags when the fall exceeds the threshold your household set.
Cover today against the number of days you said you wanted.
Flags when today's figure sits below that duration.
Cover down across consecutive snapshots with no purchase recorded. Tells a store being eaten from a store being rotated.
Flags a run of falling snapshots that nothing bought explains.
Lots inside their rotation window, first expired first out. Ninety days before a best-before date by default, thirty before a use-by.
Lists the lots to use next, with the date and its type beside each.
A lot passed its date with no consumption logged. The behavioural check, and the one that catches the exact failure the product exists to fix.
Flags the lot rather than deleting it; you may still own it.
A replacement was queued for a rotated lot and never confirmed as bought.
Flags that cover is about to drop by the amount that lot supplied.
A ration category with no representation at all: no fats, no salt, nothing carrying vitamin C.
Reports the gap as an observation about the store, not as dietary advice.
Protein cover materially below energy cover, as arithmetic over the labelled values.
Flags when the two figures diverge past a set margin.
A large share of cover sitting in one product. A store of four hundred identical tins is a store people stop eating.
Flags when one product carries more than its share of your cover.
The group that admits the picture may be wrong.
Lots dated by estimate rather than by a read or confirmed date.
Flags items whose date could not be read with confidence, so you can check the label.
Same product, same date, added twice. Re-photographing a shelf is common, and it silently doubles apparent cover.
Flags the pair for you to keep or merge; nothing is removed on its own.
Nothing logged for a period, so the inventory has probably drifted from the cupboard.
Flags that the picture may be out of date, and suggests a quick recount.
Your photographs
Four claims, each with the mechanism that makes it true. The user manual holds the long form.
Each claim is one you can check. The same four are on your photographs page once you have an account, each with where the product stands today. What happens to your photographs.
Start
Photograph one tin and you have a cover figure. Nothing else is needed to begin.