From bakery pickup to an enterprise canteen platform

ToOrdr is a German food ordering business. When the client came to us in 2019 the idea was narrow and clear: bakeries sell out of the good things early, and customers wanted a way to claim an order before making the trip. No marketplace, no delivery fleet — just a reliable reservation between a person and their local bakery.

This was the first product LexicMedia ever shipped. We were a young company being trusted with someone else’s business, and that shaped how we have worked ever since. The client gave us room to make suggestions and has never dismissed one out of hand, which is why the product has been able to change direction three times without ever being thrown away and restarted.

A pickup app for a bakery and an operations platform for a corporate canteen are not the same product, and the distance between them is larger than it looks.

In the first version an order was one customer collecting one bag at a known time. As the business moved into canteens, an order became a recurring meal entitlement attached to an employee, paid for by an employer who is not the person eating, against a menu that changes by site and by day. Headcount has to be forecast so kitchens do not over-produce. Third-party companies offering lunches to their staff need their own billing relationship, separate from the individual.

The usual answer at that point is a rewrite. A rewrite would have meant an outage in a business with customers depending on it every working day, and it would have thrown away six years of accumulated correctness — the small operational rules that only reveal themselves in production.

We grew the platform in place instead, in three deliberate generations.

Version 1 shipped the pickup flow to both app stores: browse a bakery, reserve what you want, collect it at a time that works. Small surface area, built to be extended rather than to be finished.

Version 2 was a complete rebuild of the user experience. The audience had widened from bakery regulars who chose the app to office staff who were simply told their canteen used it. Those are different users with different tolerance for friction, and an interface designed for the first group quietly fails the second. We redesigned around the new majority without breaking the original flow.

Version 3, in progress now, is adding AI-assisted automation to the operations side: demand forecasting so kitchens produce closer to what will actually be eaten, and automation of the repetitive scheduling that canteen managers were doing by hand. The client’s instinct has always been to stay ahead of the category rather than catch up to it, which is what makes this version possible at all.

ToOrdr has been continuously live since 2019 and is on its third generation. It now supports canteen operations and corporate lunch agreements alongside the original consumer pickup flow, on iOS and Android, with a backend that has absorbed three changes of business model without a rewrite.

The apps have passed 13,500 downloads — around 11,500 on iOS and 2,000 on Android. That split is worth reading carefully: this is a German user base that skews heavily to iOS, and it is the reason we invested in native Swift alongside the shared Flutter codebase rather than treating both platforms as interchangeable. Where an audience concentrates, the platform it concentrates on deserves the deeper work.

Six years is a long time in software. Most agency relationships end at launch, or at the first disagreement about scope. This one has survived three product generations and a complete change in who the product is for.

We still owe this client a great deal. They trusted a new company with the thing their business depended on, and that trust is the reason LexicMedia exists in the form it does.

ToOrdr is an iOS and Android build with a Node.js and Laravel backend behind it. Read more about how we approach mobile app development, or see the rest of our work.