Viston helped us reduce fuel costs and improve on-time delivery by implementing AI-powered demand forecasting and intelligent route optimization across our logistics operations.
Business Challenge
On-time delivery performance was at 82%, demand planning was highly volatile, and rising fuel costs were compounded by static, dispatcher-driven routing.
Our Approach and Solution
Viston developed SKU- and distribution center (DC)-level demand forecasting models along with a Vehicle Routing Problem (VRP) optimizer that considered delivery time windows, vehicle capacity, and driver constraints. The solution continuously re-optimized routes using live telemetry and real-time traffic data.
AI Demand Forecasting
Built AI-powered forecasting models to accurately predict SKU demand at the distribution center level, enabling more efficient inventory and logistics planning.
Intelligent Route Optimization
Implemented a Vehicle Routing Problem (VRP) optimization engine with real-time traffic awareness, dynamic route re-optimization, and operational constraints.
Driver Mobile Integration
Integrated driver applications with live ETA updates, exception reporting, and geofenced proof-of-delivery capabilities.
AI Applications and Benefits Delivered
Demand Forecasting
AI-powered time-series forecasting to improve inventory planning and shipment scheduling.
Route Optimization
Reinforcement-informed routing heuristics for efficient route planning and reduced transportation costs.
ETA Prediction
Accurate Estimated Time of Arrival (ETA) prediction using live traffic and telemetry data.
Business Benefits
- Reduced empty miles
- Improved On-Time In-Full (OTIF) performance
- Lower fuel consumption
- Reduced CO₂ emissions
- Increased operational efficiency
Cost of Implementation
USD 220,000, including data engineering, MLOps implementation, and mobile application integration.
Time to Implement
12 weeks for deployment across the first eight distribution depots, followed by a nationwide rollout over the next six weeks.
Tools and Technologies Used
- Databricks
- Apache Spark
- PyTorch Temporal Fusion Transformer
- Facebook Prophet
- Google OR-Tools
- Google Maps Platform
- Apache Airflow
- PostgreSQL
- Apache Superset
- MQTT Telemetry
Quantitative Outcomes
- 17% reduction in fuel cost per kilometer
- On-time delivery improved to 95%
- 14% reduction in empty miles
- 12% reduction in CO₂ emissions
- 2.4× return on investment (ROI) within 12 months
Key Performance Indicators (KPIs) Tracked
- On-Time Delivery (OTD)
- On-Time In-Full (OTIF)
- Fuel consumption per kilometer
- Kilometers per stop
- Demand forecast MAPE
- Route adherence
- Average delivery delay
Pre- and Post-Implementation Metrics
- On-Time Delivery (OTD): 82% → 95%
- Fuel Cost per Kilometer: 17% reduction
- Empty Miles: 14% reduction
- Demand Forecast MAPE: 28% → 14%
Stakeholder Quotes
Chief Operating Officer (COO)
“Dispatch stopped being guesswork. The optimizer reacts to reality, not yesterday’s plan.”
Regulatory and Compliance Considerations
- Brazil LGPD compliance
- Telematics data minimization
- Data retention policies
- ISO 14001 emissions tracking
