How a mobile app for the agricultural sector solves the problem of no network connectivity
An agronomist stands in the middle of a field in Krasnodar Krai — the smartphone catches one bar of signal, GPS jumps 10–15 meters. They need to record an anomaly on a specific plot, take a photo of the disease spot, attach coordinates, and send it to the monitoring system. In half an hour, they'll be in an area with no connection at all. A classic mobile app won't work here. We develop AgriTech solutions that withstand such conditions. Our team has 10+ years of experience in mobile development for the agricultural sector and has completed over 50 projects for farms ranging from 100 ha to 25,000 ha. Savings on logistics thanks to accurate data can reach 20–30%, and the investment in the app pays back within one season.
Our AgriTech mobile app development services include offline maps for agronomy, GPS accuracy for agriculture, ML disease recognition for plants, John Deere API integration, field tasks geotagging, crop monitoring, offline-first architecture, smart farming, and precision agriculture. We provide comprehensive solutions for agribusiness.
How AgriTech Mobile Apps Ensure Offline Map Functionality for Agronomy
AgriTech apps operate in environments where the internet is often unavailable for several hours at a time. Fields don't change every day — you can (and should) preload raster or vector tiles onto the device.
Satellite imagery. For an agronomist, Sentinel-2 or DigitalGlobe is more important than road maps. A tile server is set up with data from Earth Engine or Copernicus Open Access Hub, tiles are cached via MBTiles format and displayed using MapLibre GL Native. Before heading out, the user downloads tiles for the desired region from the last 2 weeks — NDVI changes, disease spots, waterlogging become visible. MapLibre GL provides smooth zoom and works offline — load speed is 3 times faster than online requests over poor networks (MapLibre GL Native).
GPS accuracy. The smartphone's built-in GPS gives 3–5 meters in an open field, up to 15 meters under cloud cover. That's insufficient for accurate plot mapping. Solutions:
- Bluetooth receiver with GNSS (Trimble R1, Bad Elf GPS+) provides sub-meter accuracy via the standard NMEA 0183 protocol
- SBAS (WAAS/EGNOS) correction — free, improves to 1–3 meters
- RTK via mobile internet (NTRIP client) — centimeter-level, requires constant network
Integration of external GPS on Android: UsbSerialForAndroid for USB-OTG or BluetoothSocket for BLE receivers. On iOS — External Accessory Framework for MFi-certified devices.
Field work and tasks
The agronomist in the field performs tasks: plot inspection, sampling, irrigation, fertilization. This is a workflow of geotagged tasks created by an agronomist-consultant in a web interface, and the field worker executes them in the mobile app.
Task status model: created → assigned → in_progress → completed. When transitioning to in_progress, we record the GPS start point; at completed, we record the point and photo. If the user closes a task without a network, the operation enters a queue (Room database, table pending_operations), which syncs when connection is available via WorkManager with NetworkConstraint.
Identifying crops and diseases via ML
Taking a photo of a plant and automatically identifying the disease is a popular AgriTech feature. We implement it via Core ML (iOS) or TensorFlow Lite (Android) with a model fine-tuned on the client's crops. The model is packaged into .mlmodel/.tflite and lives on the device — works offline, critical for the field.
Base open-source models: PlantNet, iNaturalist API, or models from the PlantVillage dataset (38 disease classes across 14 crops). For production, we fine-tune on the client's regional data via Transfer Learning on top of MobileNetV3.
Limitation: the model recognizes foliar diseases but not micronutrient stress — for that, multispectral drone imagery is required, not a smartphone photo.
Integration with smart sensors and machinery
The AgriTech ecosystem includes soil sensors (Davis Instruments, Decagon), weather stations, smart irrigation systems (Netafim, Lindsay), and precision agricultural machinery (John Deere Operations Center API, AGCO Connect). The agronomist's mobile app is the integration point where data from all sources converges into a single field context.
For Bluetooth sensors on iOS: CoreBluetooth with CBCentralManager, manufacturer GATT profile. For sensors with WiFi/LoRaWAN — data is aggregated by the server, the mobile client queries via REST or WebSocket.
John Deere Operations Center provides an OAuth 2.0 REST API with data on machines, fields, and operations. Integration requires a developer partner account.
Field maps and polygons
Fields are drawn as polygons: the agronomist walks the boundary or traces it on the map with a finger. Polygon input via tapping points on the map — MKPolygon / GmsPolygon with live-preview edge. We save in GeoJSON, compute area via ST_Area(ST_Transform(geom, 32637)) on the server (UTM projection for metric accuracy).
| Function | Technology | Efficiency |
|---|---|---|
| Offline maps | MapLibre GL + MBTiles | 3x faster loading in poor network |
| Satellite imagery | Sentinel Hub API / Earth Engine | Up-to-date to 1 day |
| ML disease detection | TFLite / Core ML | 85–92% accuracy on foliar pathologies |
| External GPS | NMEA 0183 / BLE | Reduces error to 1 m |
| Field tasks | Room + WorkManager | Sync on first connection |
| Integration type | Connection method | Implementation time |
|---|---|---|
| John Deere API | OAuth 2.0 REST | 2–4 weeks |
| Soil sensors (BLE) | CoreBluetooth / UsbSerial | 1–2 weeks |
| LoRaWAN | REST proxy | 3–4 weeks |
What's included in our work
With each project we deliver:
- Architectural documentation (ERD, component diagrams)
- Source code in a private repository (Git)
- Integration with App Store Connect and Google Play Console
- Access to a test environment (TestFlight / Firebase App Distribution)
- 2-day training for the client's team
- 3-month warranty on critical bug fixes after delivery
Approximate project budget
The cost of an MVP (field maps, tasks, photo capture) starts from $20,000. A full platform with ML and machinery integration ranges from $50,000 to $100,000. The exact amount is calculated after an audit of needs.Why choose us?
10+ years of experience, 50+ implemented projects in the agricultural sector, Apple and Google certifications. We know the App Store Review Guidelines (sections 4.2 and 5.1), set up code signing and provisioning profiles, work with push notifications (APNs/FCM). We guarantee stable operation of the app in field conditions. Get a consultation — we'll tell you how our solution can save up to 30% of agronomic time.
Stages and timelines
- Audit: which sensors/machinery are already in use, what GNSS accuracy is needed
- Development of an offline data model — fields, tasks, seasons
- Map and polygon handling
- ML module — dataset preparation or use of ready-made models
- Integrations with external systems
- Pilot on one farm before production
MVP (field maps, tasks, photo capture): 8–12 weeks. Full AgriTech platform with ML, machinery and sensor integration: 5–8 months. Cost is calculated individually.
Contact us for a consultation on your project — we'll select the optimal solution for your budget and timeline.