Building a multi-application delivery ecosystem

Client: Internal product — LieferXXL. Industry: Delivery and logistics.

A delivery platform needs three different applications that each serve a different audience with different priorities: customers ordering food, restaurants managing incoming orders and menus, and drivers accepting and completing deliveries. Building these as three disconnected products risks duplicated logic, inconsistent order states, and a support burden every time the definition of an order changes.

The challenge was architectural before it was visual: define a single source of truth for an order's lifecycle that all three applications could rely on, without forcing any of the three interfaces to compromise on the workflow their specific users need.

The three applications have materially different requirements: the customer app needs to be fast and simple on mobile networks of varying quality, the restaurant application needs a dependable, always-on view of incoming orders during service hours, and the driver application needs reliable location handling and offline tolerance for patchy connectivity. A shared backend needed to serve all three without becoming a bottleneck for any of them.

The order lifecycle was modelled as a single state machine shared across all three applications, with clearly defined transitions (placed, accepted, preparing, ready, picked up, delivered, cancelled) and explicit rules for which role can trigger which transition. Each application subscribes to the subset of state relevant to its users rather than duplicating the logic locally.

Shared TypeScript types and validation schemas are used across the customer, restaurant, and driver codebases so that a change to the order model is enforced at compile time across all three, rather than discovered at runtime in production.