AGRITECH / AGRICULTURAL TECHNOLOGY PLATFORM
PROPELOO engineers agritech platforms — from farm management systems and crop monitoring through IoT sensor integration, satellite imagery analysis, supply chain traceability and the offline-capable mobile apps that work without connectivity. Agricultural technology must work where agriculture happens.
Agritech has a unique constraint that most software does not: users are often in fields, greenhouses and storage facilities where connectivity is intermittent or absent. Software that fails when offline is software that fails when most needed. A farm manager recording crop observations needs that data captured immediately, even if it syncs later. A supply chain traceability system must scan produce at harvest, not only at the first point with connectivity. PROPELOO builds agritech with offline-first architecture as a baseline requirement, not a post-launch enhancement.
Field mapping, crop rotation planning, input logging (seeds, fertiliser, pesticides), labour management, equipment records and integration with accounting systems.
Satellite imagery (NDVI, crop health indices) via Sentinel-2 or Planet APIs, drone imagery processing, pest/disease identification via image classification ML models.
Soil moisture, temperature, pH sensor data collection (LoRaWAN, NB-IoT, cellular), weather station integration, automated irrigation triggering and historical sensor analytics.
Blockchain or database-backed supply chain traceability — harvest to shelf tracking, QR code generation for consumer transparency, certification documentation management and recall response.
React Native mobile app with local SQLite database for offline operation, background sync when connectivity returns, conflict resolution for concurrent edits and field-optimised UX.
Yield per hectare by variety and field, input cost per unit of output, labour efficiency, crop performance benchmarking across farms or seasons.
Agricultural software has two user populations: the office user who needs dashboards and reports, and the field user who needs something that works with one hand, in sunlight, without connectivity. Design for both.
Connectivity in agricultural settings is unreliable. An offline-first architecture assumes the device is offline by default and syncs opportunistically. SQLite on device, background sync with conflict resolution, and UI that never shows "connecting..." or blocks user actions on connectivity are the minimum bar for field software.
Axiom:
Sentinel-2 provides free 10m resolution multispectral imagery globally every 5-10 days. NDVI (Normalized Difference Vegetation Index) from satellite data shows crop health patterns across entire fields without requiring ground sensors. Building crop monitoring without satellite imagery integration in 2024 is ignoring a free, high-quality data source.
Axiom:
Farmers using software in the field are wearing gloves, working with one hand, and looking at screens in direct sunlight. Touch targets must be large, contrast must be high, form fields must be minimal, and the most common action must be the easiest to perform. Lab-tested UI designed for a desktop browser is not field software.
Axiom:
Traceability systems that require retrospective data entry produce inaccurate records. Traceability data — what was applied, when, to which field, by whom — must be captured at the point of activity, by the person doing it. This requires mobile-first capture with minimal friction and incentives for accurate recording.
Axiom:
Impact: React Native for most agritech field apps — one codebase for iOS and Android, strong offline capabilities via SQLite/MMKV, device API access (camera, GPS, Bluetooth for sensors).
Impact: Custom merge logic per data type is most practical: most agritech data has clear ownership (field observation by this worker on this day), making last-write-wins acceptable with field-level locking.
Impact: Sentinel-2 for most agritech platforms — free, global coverage, sufficient resolution for field-level crop health monitoring. Planet for applications requiring daily updates. Custom drone for very high resolution needs.
Impact: LoRaWAN for fixed sensor networks (soil, weather) — low power, long range, self-managed gateway. Cellular for equipment tracking and high-bandwidth requirements. Satellite IoT for very remote areas without LoRaWAN coverage.
Impact: Immutable append-only database for most agritech traceability — sufficient for regulatory requirements (GAP, GlobalG.A.P, FSMA) without blockchain complexity. Public blockchain only when cross-organisation trustless verification is genuinely required.
Impact: Embedded charts for operational dashboards, Metabase for ad-hoc reporting and management dashboards. AI-generated narrative insights are high-value for farmers who want "what should I do?" rather than raw data.
Field records, crop planning, input logging, labour management, compliance reporting and farm-to-accountant integration.
Satellite NDVI monitoring, historical crop health trends, field comparison dashboard and alert system for crop stress detection.
Harvest-to-shelf tracking with QR codes, retailer portal for supply chain visibility, certification document management and recall response workflow.
Soil sensor and weather station data collection, automated irrigation control, threshold alerting and historical analytics.
B2B produce marketplace connecting farmers directly to buyers — listing, price discovery, order management and logistics coordination.
Variable rate application maps from satellite data, prescription generation for fertiliser/pesticide, equipment integration and field-level ROI tracking.
Stack: React Native, Expo, WatermelonDB / SQLite (offline), Background sync, Mapbox (field mapping)
Stack: Sentinel Hub API, Google Earth Engine, Planet API, Rasterio (processing), NDVI calculation
Stack: LoRaWAN (ChirpStack), AWS IoT Core, MQTT, TimescaleDB (sensor data), Node-RED
Stack: Node.js / Python, PostgreSQL + PostGIS, Redis, AWS (primary cloud)
Stack: QR code generation, Blockchain (Hyperledger if needed), Document management, Integration APIs
Stack: Metabase, Custom dashboards (React), dbt (data transformation), ML models (Python)
Farm data belongs to the farmer. Privacy policies must be clear about data usage, sharing with third parties (input suppliers, lenders) requires explicit consent, and farmers must be able to export and delete their data.
Mobile devices used in the field may be lost or shared. Device-level PIN/biometric lock, encrypted local database, remote wipe capability.
Field sensors are physically accessible and may be tampered with. Device authentication, encrypted data transmission, anomaly detection for unusual sensor readings.
Traceability data must be tamper-evident. Append-only records, audit trail for all modifications, digital signatures for key events (harvest, certification).
Integrations with equipment manufacturers, retailers and certification bodies require proper API authentication, scoped access and audit logging.
Labour management data (worker records, hours, location) is personal data subject to GDPR. Consent, access controls and retention limits required.
Field visits with farmers and farm workers — understand actual workflows, connectivity conditions, device capabilities.
Offline-first mobile design, data model, sync architecture, IoT connectivity plan.
Field data capture, offline operation, GPS-tagged records, camera integration.
API, database, sync engine, conflict resolution, web portal.
Satellite imagery, IoT sensors, supply chain partners, accounting software.
Farm performance dashboards, trend analysis, benchmarking.
Testing in actual field conditions — connectivity gaps, sunlight readability, glove use.
Most agritech data has natural ownership that prevents conflicts: one field observation per worker per field per day, one harvest record per field per date. Conflict prevention via data ownership rules eliminates most merge scenarios. For the remaining cases: last-write-wins with a conflict log is practical. True CRDTs are overkill for most agritech data models.
NDVI (Normalized Difference Vegetation Index) is calculated from near-infrared and red light bands in satellite imagery. Healthy vegetation absorbs red light and reflects near-infrared; stressed vegetation does not. NDVI values range from -1 to 1 — values above 0.3 indicate healthy vegetation. NDVI maps from Sentinel-2 satellite data show field-level crop health patterns, identifying problem areas before they are visible to the naked eye.
A practical farm-to-shelf traceability system works as follows: field harvest is recorded in the mobile app (crop, field, date, quantity, lot number). Each lot gets a unique ID and QR code. At each subsequent step (packing, cold storage, transport, retail), the QR is scanned and the step is recorded. A consumer or retailer scanning the final product QR sees the complete chain. Blockchain adds tamper-resistance; a well-designed append-only database provides sufficient integrity for most food safety regulatory requirements.