Customer app
Browse, order, pay and watch the rider on the map.
Case studies · Food & delivery
Bite — food delivery across Lahore
Three published apps over one platform — customer, restaurant and courier — serving 3,000+ restaurants and stores in Lahore, with live rider tracking, a wallet and scheduled orders.
3Apps in the stores
10K+Installs, customer app
3,000+Restaurants and stores
3Months, scope to launch
The problem
Where it started.
A delivery marketplace is not one product. It is three apps and an admin console that have to agree on the state of a single order at the same second — the customer watching a map, the rider accepting the job, the kitchen confirming it can cook it, and the office setting what it costs. Get that disagreement wrong and every side loses trust at once: the customer sees a stale map, the rider drives to a closed kitchen, the merchant is paid for an order that was refunded.
Bite had to launch into Lahore against incumbents who already had the restaurants and the riders, which meant the ordering experience could not be a rough first version — it had to be right on day one. All four surfaces, the core that runs them and the payment rails underneath went from scope to the stores in three months — and we drew every screen in all three apps as well as building them.
Architecture
One core.
Four surfaces over it.
Every rule about an order lives in one place, so a change to pricing or availability never has to be made four times.
Core
API
API
Vendor app
Menus, availability, prep times and order acceptance.
Rider app
Shifts, job acceptance, routing and delivery status.
Admin console
Zones, fixed delivery charges, promotions, support.
In the stores
Three apps.
One order between them.
Each side of the marketplace got its own app, designed and built by the same team so the three never disagree about the state of an order. Install figures and dates are what the listings showed on August 13, 2026.
Customer app
Bite
Browse 3,000+ restaurants and stores, order, pay by wallet, card or cash, and watch the rider move across a map of Lahore.
- Installs
- 10K+
- Category
- Food & Drink
- Last update
- July 2026
Restaurant app
bite Vendor
Accept and track orders, edit menus, prices and availability, watch stock, and read the payout and performance dashboard.
- Installs
- 500+
- Category
- Food & Drink
- Last update
- March 2026
Courier app
Bite Rider
Start and end a shift, take jobs in a zone, navigate the drop and keep a delivery history that survives a dead signal.
- Installs
- 100+
- Category
- Maps & Navigation
- Last update
- August 2026
Scope
Everything it had to cover.
Three apps and an admin console over one core, grouped the way the business actually runs rather than the way the code is organized.
Ordering
- Browse by cuisine, rating, delivery time or distance
- Restaurants, pharmacy and pick-up as separate verticals
- Item customization and per-item special instructions
- Schedule an order for later, or reorder from history in one tap
- Favorites, ratings and reviews
Delivery
- Live GPS tracking from kitchen to door
- Rider position on the map, pushed rather than polled
- Dispatch, acceptance and routing on the rider app
- Contactless handover and proof of delivery
- Fixed delivery charges, shown before checkout
Money
- Wallet top-up and running balance
- Card and cash alongside the wallet
- Refunds and order-level adjustments
- Merchant payouts and settlement
- Commission and fee rules held centrally
Operations
- Delivery zones drawn and priced per area
- Menus, prep times and availability held by merchants
- Promotions and voucher campaigns
- Dispute queue for support to work from
- Push notifications on every order state change
Engineering
The hard parts.
Any team can list features. These are the problems that decided whether the platform worked at all.
One order, four clients
Every rule about an order — pricing, availability, who may change what and when — lives in a single state machine on the core API. The apps render state, they do not own it. That is the only reason four applications could be built in parallel in three months without drifting apart.
Real-time location, cheaply
The customer sees the rider move without hammering the network or the battery. Position is streamed and pushed to the order channel rather than polled, which also keeps the map honest when a rider drops signal in a basement or a car park.
Geography as a first-class concept
Delivery zones, distance filtering, and which restaurants can even appear for an address are geospatial questions answered per request, across a city the size of Lahore, fast enough that browsing never waits on them.
Money that reconciles
A wallet balance sitting alongside cash and card means every order has to reconcile three ways: what the customer paid, what the rider collected, what the merchant is owed. Refunds and adjustments have to survive that without leaking.
Built for a bad connection
Riders work in lifts, basements and traffic. Actions queue and reconcile rather than failing, because a job accepted on a dead signal still has to be a job accepted.
Release discipline
Weekly TestFlight builds on real phones in Lahore throughout, then staged App Store releases — live in September 2025 and still shipping a year later.
Gallery
A look at it.
01/ 03
01
Customer app
Bite
Browse Lahore, order in a few taps, and watch the rider close the distance on a live map.
02
Restaurant app
bite Vendor
Accept the ticket, change the menu, watch stock and read the payout — from the counter, on a phone.
03
Courier app
Bite Rider
Start a shift, take a job in your zone, navigate the drop, and keep a history that survives a dead signal.
Process
How we ran it.
- 01
Modelled the order first
One state machine for the order, agreed before any screen was drawn.
- 02
Built the core, then the surfaces
API, auth and payments ahead of the apps, so each surface stayed thin.
- 03
Shipped to TestFlight weekly
Real builds on real phones in Lahore throughout, not a reveal at the end.
- 04
Launched and kept shipping
Live in September 2025 and still shipping — all three apps updated again in 2026.
Outcome
What changed.
3,000+Restaurants and stores on the platform
10K+Installs of the customer app on Google Play
2,000Orders a week at steady state
30 minAverage delivery time, end to end
They did not treat it as three apps and a website. They treated it as one system with three doors into it, and that is the only reason the customer, the restaurant and the rider have never disagreed about an order. We went from a plan to a working platform in three months.
Daud, CEO, Bite Pakistan
Stack
Built with.
- React Native
- Laravel
- MySQL
- Redis
- Pusher
- Firebase
- Google Maps
- Push notifications
Got a problem
shaped like this one?
Thirty minutes on your process. We'll tell you what we'd build first and what we'd leave alone.
Book the call→Or email hello@devjunction.co