Inspiration
arabic-first delivery app for West Bank cities.based on uber and karem apps for inspiration
self-service accepts name, email, Palestinian mobile number and password. Customers are signed in immediately; driver applicants select a vehicle and wait for administrator approval. Admin accounts cannot be self-created. Contact values are unverified until a real verification provider is added. Administration: an administrator can search, approve, block, edit or delete customer and driver accounts. Identity, role and status changes revoke sessions. Driver GPS: a driver enables foreground location before accepting a job. Wind starts a high-accuracy point immediately after acceptance and labels sharing only after the API receives it; collection stops when the active job or foreground app ends. Vehicle matching: motorcycles can take parcels and licensed rides; bicycles are parcel-only. The database upgrade preserves existing user, session, order and GPS data. Connection to Souq: the two apps keep separate users, sessions and payments. Their delivery bridge is private server-to-server traffic. The final device, API, signed-artifact and backup checks are recorded in ../Deployment-2026-10-03/.
What it does
Customer: book, see the price before confirming, track the driver live, hand over with a 4-digit code, rate, Wind Plus plan. Driver: register with a motorcycle, bicycle, car, van or truck, wait for approval, accept matching jobs, see the latest completed trip and earnings, Wind Pro plan. Admin: revenue dashboard, all orders, approve/block users, reset passwords, reopen delivery-code attempts after recipient verification, activate plans, edit tariffs and commissions.
How we built it
One Expo app serves for iphone and android , we built it in usign azure infrastructure and we use Ai model for coding and reviewing the app includeing gemini and claude and open ai models and self human reviews , the app now is in the production
Challenges we ran into
the cost of hosting services and ai model for coding cost , we struggle with maps and users tracking in the app maps and how we can make low cost map and gps tracking and deliver best quality to our users
Accomplishments that we're proud of
We studied a production ride-hailing/freight app's architecture and shipped the rides+freight slice into Wind with full test evidence ,including a real on-device-discovered crew bug
What we learned
The big lesson was verification discipline. Early on, we had documents confidently stating things that turned out wrong once we double-checked them, so we built a habit of re-verifying everything against the actual source code before trusting it — and where a document and the code disagreed, the code won.
Second, we learned that we'd already built far more on the backend than the app was actually using. The gap wasn't architecture, it was wiring and tests. So the rule became: extend what exists, don't rebuild it.
Third, testing on a real device caught real bugs that the automated suites missed. And when we deliberately broke the code to see if the tests would catch it, they did — which is how you know your testing means something.
And honestly, the clearest thing we learned is precision about what you don't know. We kept an explicit list of what was unproven — other platforms, accessibility, real-world GPS conditions — instead of claiming victory early
What's next for wind
Near term, it's about closing the gap between 'works in test' and 'ready for real customers.' That means applying the database changes we built, finishing a controlled pilot with actual drivers and real operating conditions, and gathering the operational evidence that proves the system behaves outside a test harness — including a rollback plan, not just a launch plan.
We also have acceptance work left on devices we haven't fully covered, and accessibility checks that were never done. Those aren't nice-to-haves; they're part of calling it done.
After that, the roadmap is about differentiating for our actual market — Arabic-first experience, payment habits that fit the local economy, tolerance for unreliable networks, and honest pricing transparency. The strategy stays the same throughout: prove each step properly before moving to the next, and be clear-eyed about what's verified versus what isn't
Built With
- amazon-web-services
- azure
- claude
- firebase
- kimi
- openai