How Appther built an end-to-end AI logistics platform for SwiftLogistics — combining Python VRP optimisation, real-time GPS telematics, and predictive ML models to transform a 120-vehicle fleet operation.
Results achieved within 90 days of go-live
SwiftLogistics operated a 120-vehicle fleet across 3 depots, handling 2,000+ deliveries daily. Dispatchers manually built routes in spreadsheets each morning — a 3-hour process that still produced inefficient paths, ignored real-time traffic, and left 40% of vehicle capacity unused.
Rising diesel prices made every wasted kilometre costly. Late deliveries damaged NPS, and competitors with digital logistics platforms were undercutting them on price.
End-to-end system covering route optimisation, fleet visibility, and driver empowerment — deployed across 3 depots in 5 months.
A Python-based optimisation service using Google OR-Tools and live traffic data to compute the shortest, cheapest multi-stop routes in under 200 ms.
A real-time React web dashboard giving dispatch teams a live map of every vehicle, ETA for every stop, fuel metrics, and one-click route reassignment.
A lightweight React Native app with turn-by-turn navigation, proof-of-delivery capture, and automated customer ETA notifications — no manual calls needed.
Considers real-time traffic, delivery time windows, vehicle capacity, and priority tiers to generate optimal routes instantly.
Live GPS pings every 10 seconds with idle-time detection, geofence alerts, and predictive maintenance scoring.
Per-route and per-driver fuel analytics with anomaly detection — flags inefficient driving and recommends vehicle reassignment.
ML-predicted delivery windows with ±8-minute accuracy and automated WhatsApp/SMS updates at key milestones.
Assigns parcels to vehicles based on weight, volume, fragility, and geographic clustering to maximise load utilisation.
End-of-day reports on cost-per-delivery, driver scores, SLA breach rates, and fuel spend — exportable to CSV or BI tools.
LSTM neural network trained on 2 years of historical order data predicts next-day delivery volumes per zone with 91% accuracy — allowing the depot to pre-position vehicles before orders arrive.
When traffic, road closures, or urgent deliveries occur mid-shift, the AI engine recalculates and pushes updated routes to affected drivers within 30 seconds — no dispatcher needed.
Analyses telematics data (RPM, braking, engine temps) to flag vehicles likely needing maintenance in the next 14 days — preventing breakdowns that previously cost 6% of fleet uptime.
From route audit to full fleet deployment — a systematic approach that eliminated risk and accelerated ROI.
PythonAI / Backend
ReactWeb Dashboard
Node.jsAPI Layer
MongoDBDatabase
AWS IoTGPS Telematics
Google Maps APIMappingReact NativeDriver AppThe Google OR-Tools VRP solver returns optimised multi-stop routes for up to 200 deliveries in under 200 milliseconds. For larger batches the engine parallelises across AWS Lambda and still returns results within 2 seconds.
Yes. We built a REST API layer specifically for third-party integration. SwiftLogistics connected their existing SAP TMS in 2 weeks with no downtime using our webhook-based connector. We support CSV import, REST, and SFTP-based integrations.
All location data is transmitted over TLS 1.3 and stored encrypted at rest in AWS with per-depot access controls. Customer PII is pseudonymised in analytics exports and the platform is GDPR-compliant by design.
The driver app caches the current route and works fully offline. Once connectivity resumes, it syncs proof-of-delivery photos, stop completions, and re-routing updates automatically — no data is lost.
Drivers are fully productive within one working shift. SwiftLogistics completed training across 120 drivers in a single day using our train-the-trainer model with the in-app guided tour.
Absolutely. The platform was built from day one for SwiftLogistics' 3-depot operation. Each depot has independent route planning with shared fleet analytics, cross-depot reassignment capability, and a unified management dashboard.
GET IN TOUCH
Share your project requirements and our team will get back to you within 24 hours.