Cut-off firming
18 choices still open
Defaults apply at 21:00 unless customers change them. The kitchen count firms up on its own.
A home-cooked meal service with a daily rotating menu, cooked by a kitchen team and delivered across multiple zones. For three years it had run on a rigid third-party platform: nothing could be changed or improved, its only data export was a restricted daily order list, the menu was published as an uploaded picture, and the business was invisible to search. Percei recovered the entire operation, rebuilt it around the kitchen’s real day, and took it live: subscriptions and one-off orders, a rotating menu with cut-off times, portion-counted prep, delivery run sheets by zone, online payments, self-serve customer accounts, and an indexable storefront. One connected system.
The operator publishes a rotating menu; customers choose, or let their default stand, before the daily cut-off. By dawn the system has already turned every confirmed choice into a kitchen manifest: dishes, portion counts, dietary flags, and a delivery run sheet ordered by zone. Nothing is tallied by hand.
A home-cooked meal service delivered a daily rotating menu to subscribers across three neighbouring zones, cooked by a small kitchen team. The software behind it was a third-party platform the business rented: it held the customers, the orders, the addresses, and the payment history, but it was built for a generic storefront, not for a kitchen with a cut-off time and a morning of prep.
And it had stopped moving. For three years no change was possible: the platform couldn’t be extended or improved, its menu was published as nothing more than an uploaded picture, and the business had no presence in search. Every day still started by reading its screens and rebuilding the picture by hand: who ordered what, how many portions of each dish, which addresses were on today’s run.
The platform did offer an export, but a heavily restricted one: little more than a daily order list. Years of customers, orders, and addresses couldn’t come out cleanly, and the software itself couldn’t be reshaped to fit how the kitchen actually worked.
And because nothing turned confirmed orders into a count the kitchen could cook to, prep was estimated and deliveries were sequenced from memory. A missed portion or a last-minute skip surfaced at the door, too late to fix cleanly, and a paused meal a customer had paid for was easy to lose track of entirely.
Defaults apply at 21:00 unless customers change them. The kitchen count firms up on its own.
The premium pick is trending. Surcharge and prep count are already reflected in the manifest.
Logged against each account and re-scheduled automatically. No paid meal is lost to a skip.
Heaviest of the zones. The run sheet is ordered by round so the driver moves in one line.
Subscription packages, a rotating menu of dishes, and published sessions with per-customer choices and cut-off times, plus one-off and seasonal orders. The model matches how the kitchen actually plans, so a changing menu never breaks an active subscription.
A daily prep sheet counts portions per dish with dietary flags, a delivery run sheet sequences drops by zone, and a skip-and-make-up ledger keeps paused meals owed and re-scheduled, so a paid meal is never quietly lost.
A self-serve portal for plans, menus, and addresses; online payment with reconciliation; and full Chinese localization throughout. It all works on a phone, so customers handle their own accounts without a call.
The most valuable part of this build was never the admin screens. It was modelling the kitchen’s real day: publish the menu, let choices lock at the cut-off, cook to an exact count, run the deliveries by zone, and settle the skips into make-up meals. Getting the data model right came first: before any of this could work, the entire history had to be recovered from the old platform and rebuilt into a schema that understood packages, sessions, and portions. From there, the software could finally run the loop instead of just recording it.
The business went live on its own domain with the full customer base migrated intact and effectively zero drift from the recovered records. Orders and payments now flow in directly; every morning the portion counts and the run sheet are already waiting. The team spends the day running the kitchen, not reconstructing it from someone else’s screens.
One flow: a customer’s choice becomes a counted kitchen operation becomes a meal at the door.
We build operational software around how your operation actually works, not the other way around.
Book a 30-minute introNo commitment. No deck. Thirty minutes is enough.