What it takes to build an e-bike app with Bluetooth, GPS and anti-theft
An e-bike app needs two links to the bike. Bluetooth Low Energy handles everything that happens within a few meters: pairing, the digital key, live ride data and settings. Anti-theft needs a second link that works when the phone is nowhere near, which means an IoT module on the bike with its own GPS, a cellular modem and a SIM. For most brands the fastest route is a third-party IoT module wired into the drive system, a React Native app on top of the vendor's SDK, and a budget of $70K to $130K over three to five months for a first version on iOS and Android.
This guide is for a founder or product lead at an e-bike brand who has a bike, a drive system supplier and a decision to make about the app. It sits one level below our general guide on building a Bluetooth app for Android and iOS hardware. Read that one for GATT design, bonding and permissions. This one covers what is specific to a bike: who owns the motor protocol, what theft recovery needs from the hardware, and why the app must never raise the speed limit.
The hard part is access. Whether you can read the battery, lock the motor or push firmware depends on a contract with your drive system supplier, and that contract should be settled before anyone designs a screen.
How the app talks to the bike
Inside an e-bike, the motor controller, the battery management system and the display talk to each other on a wired bus, usually CAN on current systems and UART on older or cheaper ones. Your phone cannot join that bus. Something on the bike has to bridge it to a radio.
Bluetooth in the display or controller
Many displays and some controllers ship with a BLE chip. The phone connects to it and reads speed, assist level, battery state and error codes, and writes settings back. It is the cheapest option because the hardware is already on the bill of materials. Bluetooth only works next to the bike, so it can do a digital key and ride stats, and it cannot tell anyone where a stolen bike is.
An IoT module with cellular and GPS
An IoT module is a small sealed board, hidden in the frame or under the motor cover, with a GNSS receiver, an LTE-M modem, an accelerometer, Bluetooth and usually a small backup battery. It listens on the CAN bus, reports to a cloud over cellular, and can tell the controller to cut motor assist. Tracking, motion alerts and remote lock all depend on it. It adds hardware cost to every bike and a data plan somebody pays for every month.
Both
The connected bikes worth copying use both paths. The phone talks to the module over Bluetooth when it is close, because that is instant and works in a basement with no signal. The cloud reaches the module over cellular when the phone is far away. Your app shows one bike with one honest status.
If you design your own controller or display firmware, you own the Bluetooth protocol and can shape it around the app. If you buy a drive system off the shelf, the protocol belongs to the supplier and you get whatever access they sell.
Bosch, Shimano, Bafang and the IoT module vendors
I looked up what each supplier offers a brand today. The closed systems give you data and keep control. The open ones give you control once you sign an agreement.
Bosch. The smart system is the most closed and the best documented. Riders use Bosch's own eBike Flow app for the lock, the alarm and updates, and your app does not replace it. What Bosch does offer is a Live Data Interface, which it describes as free on-bike read access over BLE to values such as speed, rider power, cadence, battery state of charge, odometer and lock state. It also lists cloud APIs for ride and profile data that the rider has to approve. On a Bosch bike, plan a brand app that sits beside Flow, with registration, service and ride history, and leave the digital key to Bosch.
Shimano. Shimano drive units are configured through its E-TUBE apps. I could not find a public Bluetooth protocol document for third parties, and Shimano's software license terms prohibit reverse engineering. Access goes through your Shimano OEM account, or through an IoT partner. Velco, a French connectivity vendor, has announced a collaboration with Shimano for bike brands.
Bafang. Bafang moved its newer systems from UART to CAN and services them with its BESST tool. I found no public protocol specification. Brands get documentation through their supply agreement, and Comodule publishes a dedicated integration for Bafang systems, which saves you months of bus sniffing.
IoT module vendors. Comodule in Estonia and Velco in France both sell the module, the connectivity, a cloud and an app layer to e-bike brands. Comodule's line includes the Astra and an entry-level module called Guardian, which the company says connects over LTE Cat-M1 with a 2G fallback. It lists a Bluetooth SDK and an API for brands that want their own app, plus a white-label app for brands that do not. SUPER73 is one US brand that has publicly shipped GPS tracking, movement alerts and a remote immobilizer on Comodule hardware.
My recommendation for a brand selling fewer than about ten thousand bikes a year: buy the module, use the vendor's SDK for the Bluetooth and cloud plumbing, and spend your money on an app that feels like your brand. Building your own telematics board means running a second hardware company inside the first, with cellular certification on top.
The features that make or break it
Pairing and the digital key
Pairing should take under a minute in a bike shop with twelve other bikes of the same model in the room. Put a QR code carrying the bike's serial on the frame or the display, so the app connects to the right one. Tie ownership to an account on your server so it survives a new phone.
The digital key is the feature riders show their friends: walk up with the phone in a pocket and the motor arms, walk away and it locks. Be clear in the copy that a motor lock makes the bike miserable to ride and hard to resell, and does nothing about two people lifting it into a van. Design the dead-phone case on day one, with a PIN on the display or an NFC card, because a rider stranded by a flat phone battery will return the bike.
Ride tracking, battery and range
Speed, distance, assist level and battery percentage come off the bus and are easy. Range is hard. It depends on assist mode, rider weight, hills, wind and temperature, and riders remember every time the number was wrong. Start with a plain estimate from the rider's own recent watt-hours per mile, show it per assist mode, and label it as an estimate.
GPS anti-theft
Theft is why people pay for a connected bike. The rider arms the bike, the module's accelerometer detects movement, the module wakes its modem and sends an event, and the rider gets a push notification within seconds. If the bike keeps moving, the module switches to a tracking mode and reports position every few seconds instead of every few hours. The app shows a live map and a theft report with the frame number that the rider can hand to the police or an insurer.
A few details decide whether a tracker recovers bikes. The module needs its own battery, because the first thing a thief removes is the main pack. GPS fails indoors, so the module should fall back to cell tower or Wi-Fi positioning and the app should show the accuracy circle. False alarms have to be rare, and a bike on a train or a car rack moves without being stolen, so let the rider snooze alerts.
Firmware updates over the air
An e-bike has several computers: the controller, the display, the battery management system and the IoT module. Each may need updates, and a failed controller update leaves a customer with a very heavy bicycle. Refuse to start below a battery threshold or while the wheels are turning. Roll out to a small share of bikes first and watch error rates before you release to everyone.
Service and warranty
Service is the quiet feature that pays for the app. Register the bike at purchase, read the odometer for service reminders, translate error codes into plain sentences, and send the dealer the bike's diagnostic snapshot before the customer arrives. Add ownership transfer for resale, or every second owner becomes a support ticket.
Background Bluetooth and location on iOS and Android
A proximity key and automatic ride recording both need the app to work while it is closed. Most of the engineering time goes here.
iOS. You declare the Bluetooth background mode, scan only for your own service UUIDs, and use Core Bluetooth state restoration so the system relaunches the app when the bonded bike comes into range. On iOS 26, the current release, Apple's guidance on restoration ties relaunch behavior to AccessorySetupKit, the system pairing sheet introduced in iOS 18. Pair the bike through AccessorySetupKit from the start. Test force-quit behavior on real phones, because developers are still reporting edge cases in Apple's forums.
Ride recording uses location in the background. Ask for location while the app is in use first, start the session when the bike connects, and only ask for Always if you have a feature that needs it.
Android. Since Android 12, Bluetooth uses the Nearby devices permissions. A long-lived connection runs in a foreground service with a visible notification, and since Android 14 that service has to declare a type, which for a bike is connected device plus location while a ride is recording. Background location is a separate permission with its own Play Console declaration. Android 17, out since June, keeps the same model. The bigger risk is manufacturers whose battery managers kill your service anyway. Budget for a spread of test phones.
The good news for anti-theft is that none of this applies. Theft alerts come from the module through your cloud as push notifications, so they arrive whether or not the app has run that week. Keep security on the cellular path and convenience on the Bluetooth path.
The stack for the app, the backend and the maps
App. React Native with an Expo development build is my default for an e-bike app. Most of the product is accounts, maps, ride history and service screens, which share cleanly across platforms. Bluetooth goes through react-native-ble-plx or through the module vendor's native SDK wrapped in a small native module. Go native in Swift and Kotlin if the proximity key is the headline feature and you own the protocol end to end, since that is where background behavior gets tuned by hand. Flutter with flutter_blue_plus works too. I pick it less often for hardware work.
Backend. You need accounts, bike ownership, ride storage, firmware hosting, notifications and a support console. A TypeScript API on Postgres with PostGIS for location queries covers it. Keep raw position history for a fixed period, then delete it. The location history of a named person is sensitive data.
Device cloud. If you use a module vendor, their cloud receives the cellular traffic and forwards events to your backend through webhooks or an API. If you build your own module, the standard pattern is MQTT into a managed broker such as AWS IoT Core, with a device registry and signed firmware delivery. That is its own workstream, closer to our smart home IoT app cost guide than to a mobile project.
Maps. Mapbox, Google Maps and Apple MapKit all handle a ride track and a live bike position. The cost question is turn-by-turn directions. Bike routing with voice guidance is a product in itself and every provider bills for it by usage. Leave it out of version one. Riders already have a maps app.
Speed classes, store review and SIM plans
Speed classes
Most US states use a three-class model. Class 1 is pedal assist up to 20 mph. Class 2 adds a throttle, also capped at 20 mph. Class 3 is pedal assist up to 28 mph. Federal consumer product law defines a low-speed electric bicycle as one with a motor under 750 watts and a top speed under 20 mph on motor power alone. In the EU, a standard pedelec is limited to 25 km/h and 250 watts of continuous power. States differ at the edges, and a few have rewritten their rules this year, so check each market you sell into.
The rule for the app is simple. It must not be able to raise the assist limit above the class on the bike's label. That means no hidden menu and no off-road mode behind a disclaimer. California went further in 2025 with AB 545, which amended Vehicle Code section 24016 to prohibit selling an application that can modify the speed capability of an electric bicycle. Your app can lower the limit, for a teenager's bike or a rental. Enforce the cap in the controller firmware so a modified app cannot bypass it.
App store review
Reviewers at Apple and Google do not own your bike. Give them a demo mode with a simulated bike, a video of pairing and the key, and login credentials for an account with ride history. Expect a question about background location on both stores and have a one-sentence answer ready.
Cellular data plans
Every tracker needs a SIM, and the SIM outlives the sale. Module vendors usually bundle connectivity into a per-bike fee. If you source your own, IoT carriers sell low-volume plans cheaply. 1NCE, for example, sells a prepaid plan covering ten years and 500 MB for a one-time fee per SIM instead of a monthly bill. Decide now who pays after the first year or two. The common answer is a period included with the bike, then a rider subscription that also covers theft features. Design the billing for it into the first release.
What an e-bike app costs and how long it takes
These are market ranges for a competent team, not a quote from us. For how app budgets break down in general, see our app development cost page.
- Companion app on a module vendor's SDK, $70K to $130K, 3 to 5 months. iOS and Android, pairing, digital key, ride history, battery and range, theft alerts with a live map, registration and service reminders. The vendor's cloud does the device work.
- Your own Bluetooth protocol and backend, $130K to $220K, 5 to 8 months. Everything above, plus firmware updates for several components, staged rollouts, a support console, subscriptions and a dealer view.
- Your own IoT module and device cloud, $220K and up, 8 months or more. Add device provisioning, an MQTT pipeline, SIM management and a test fleet. The hardware, certification and firmware budgets are separate and usually larger.
What moves the number: how much of the protocol is finished when the app team starts, whether the proximity key has to work with the app closed, how many bike models and firmware versions you support, and whether theft features sit behind a subscription.
After launch, plan on 15 to 25 percent of the build cost per year for OS updates, new phone testing and firmware compatibility. Per bike, connectivity and cloud usually land at a dollar or two a month, more if riders use live tracking often.
How we would build it
We have not shipped an e-bike app, so here is how we would scope one, based on how we build consumer apps. The closest proof of pace we can point to is Vybes, a dating app with no hardware in it at all, which went from concept to the App Store in 8 weeks and holds a 4.8 App Store rating. A bike adds firmware and a supplier to that schedule, which is why the first two weeks below are spent on a bench.
The first version
One bike, one module, one iPhone, one Android phone and a plain test app. We would pair, read the battery, arm the lock, pull the main pack, carry the bike out of the building and time the push notification. If any of that fails, the plan changes before design starts.
Then the smallest app a rider would keep: QR pairing through AccessorySetupKit on iOS and the Nearby devices flow on Android, a digital key with a PIN fallback, a home screen with battery and an honest range estimate, ride history, theft alerts with a live map, and bike registration. The speed cap lives in firmware and the app has no setting that touches it. Builds go to TestFlight and Play internal testing every week, on real bikes ridden by your own staff.
What comes after
Firmware updates over the air come second, as soon as the first bikes are in customers' hands. Then the subscription for theft features once the included period is defined, the dealer and service tools, ownership transfer, and range estimates trained on a season of real rides. Turn-by-turn directions and social features go last, if they go in at all.
If you have a bike, a drive system supplier and a launch date, bring us the protocol documents and the module you are considering, and we will tell you what the first version takes. We build apps like this. Book a free strategy call


