4
Surfaces
customer, driver, owner dashboard, POS
saas · 2026
A complete restaurant operating system on one multi-tenant backend: customer ordering, driver dispatch, owner dashboard and point of sale.
4
Surfaces
customer, driver, owner dashboard, POS
51
Dashboard pages
58k
Frontend code
lines across the surfaces
Multi-tenant
Tenancy
scoped at the data layer
The problem
Independent restaurants take orders over messaging apps and track them on paper. Delivery platforms charge heavy commission and keep the customer relationship, so the restaurant never learns who its regulars are. Owners have no live view of orders, drivers or daily revenue, and no system that covers both delivery and walk-in trade.
The approach
Built four surfaces over one multi-tenant backend where each restaurant is an isolated tenant sharing a single codebase and deployment. A customer ordering experience for browsing and checkout, a driver surface for assignment and live location, an owner dashboard covering menu, orders, drivers and analytics, and a point of sale for counter trade so both channels land in the same order model. Tenant scoping is enforced at the data access layer rather than trusted to individual queries.
Architecture
One API with tenant-scoped data access enforced centrally. Live order and driver events over per-tenant socket rooms. Delivery and counter orders share one order model so reporting and stock behave identically regardless of channel.
Each with what it cost
A decision without a stated tradeoff is a preference.
01
Why
One cluster, one deployment, one migration path. Onboarding a restaurant is a row rather than an infrastructure task.
What it costs
Every query has to be tenant-scoped, so that scoping lives in one data access layer where a missed filter is impossible rather than in every controller where it is inevitable.
02
Why
Reporting, refunds, stock effects and customer history are identical logic regardless of how the order arrived. Two models means writing all of it twice and watching them drift.
What it costs
A few channel-specific fields are nullable on the shared model, a far smaller cost than duplicated business logic.
03
Why
Zero install friction for customers and drivers, one codebase per surface, and updates that ship without an app store review.
What it costs
Push notification support is weaker than native, handled through web push.
Outcome
A restaurant goes from signup to taking online orders in under a day, keeps its full margin, and owns its customer data. Four surfaces across roughly 58,000 lines of frontend and a shared backend, with 51 dashboard pages and 27 customer pages.