A technical founder asked us to rebuild his delivery app. We told him not to.
The call
He came to us with a working idea and a working app he'd outgrown. He could see the months of rebuild ahead of him — five clients (customer, rider, vendor, admin, web) wired for a global, multi-currency, multi-payment, social-login world he was never going to need. He wanted us to start over and build something native to his market.
We told him there was a faster way. Keep Enatega as the client surface — it's an open multi-vendor codebase that ships those five apps already. Let us build the backend it needs to actually run in Mauritania.
What we did
We collapsed the generic stack down to the specific reality. One currency (MRU). One country code (+222, locked at the API boundary so non-Mauritanian numbers can't enter). One auth flow — WhatsApp OTP, with the Google/Apple/Facebook logins removed entirely. One payment — cash on delivery, enforced server-side. One language priority — French first, Arabic and English alongside it. Realistic seed data for Nouakchott so the first launch isn't an empty screen.
Every order event flows through WhatsApp. Pending, accepted, picked up, delivered — the customer, vendor and rider each get a message at the moment that matters. The same channel handles sign-in OTP, replacing both Firebase push and email.
Along the way we built a small project-scoped agent that owns one rule — adapt the API, not the frontend — so every contract mismatch that used to cost engineering time now resolves itself in a single round-trip.
Where it sits now
A live operation, in a fraction of the months he'd budgeted, on a codebase he can still upgrade. And a relationship that started with us saying no to the bigger invoice.
NestJS, GraphQL, Postgres, Expo SDK 54, WhatsApp via WahaSend, Expo Push.
