We collaborated with Viston to optimize routing, dispatch, and driver support by implementing an AI-powered operations copilot in Brazilian Portuguese.

Business Challenge

Rising last-mile delivery costs, inconsistent on-time delivery performance, and increasing dispatcher workload.

Urban traffic congestion, restricted delivery zones, and seasonal demand spikes created additional operational challenges.

Our Approach and Solution

Built a daily and intra-day route re-optimization engine supported by an AI dispatcher copilot.

Deployed a driver mobile assistant to manage delivery exceptions, proof-of-delivery, and real-time micro-route adjustments.

AI Applications and Benefits Delivered

Route Optimization

Operations Research (OR)-based route optimization with live telemetry feedback loops to improve delivery efficiency.

ETA Prediction

AI-powered Estimated Time of Arrival (ETA) prediction with dynamic route re-sequencing based on real-time traffic conditions.

AI Operations Copilot

LLM-powered dispatcher copilot that summarizes delivery exceptions, drafts customer communications, and provides multilingual translations with tone control.

Cost of Implementation

USD 90,000, including telematics integrations and mobile application enhancements.

Time to Implement

16 weeks for the Minimum Viable Product (MVP) across two cities, followed by a 28-week nationwide rollout.

Tools and Technologies Used

• Google OR-Tools
• PyTorch
• PostgreSQL
• PostGIS
• Apache Kafka
• FastAPI
• Flutter
• Mapbox
• OSRM
• Redis
• LangChain
• Azure OpenAI Service (or local Llama 3 as fallback)
• Grafana
• Prometheus
• Sentry

Quantitative Outcomes

• 17% reduction in cost per delivery.
• 11 percentage point improvement in on-time delivery.
• 12% reduction in CO₂ emissions per package.
• 23% fewer dispatcher interventions.

Key Performance Indicators (KPIs) Tracked

• Cost per delivery
• On-Time In-Full (OTIF)
• Average route duration
• First-attempt delivery success rate
• ETA Mean Absolute Error (MAE)
• CO₂ emissions per package
• Dispatcher interactions per 100 orders

Pre- and Post-Implementation Metrics

• Cost per Delivery: BRL 16.40 → BRL 13.60
• On-Time In-Full (OTIF): 82% → 93%
• ETA Mean Absolute Error (MAE): 14.2 minutes → 7.6 minutes
• Dispatcher Interactions per 100 Orders: 61 → 47

Stakeholder Quotes

Head of Operations

“The copilot shaved minutes off every exception and standardized customer communications.”

Regional Logistics Manager

“Drivers trust the app’s re-sequencing when traffic spikes.”

Regulatory and Compliance Considerations

• Brazil’s LGPD compliance
• Telemetry data minimization
• Opt-in consent for driver analytics
• Anonymized operational reporting
• Compliance with ANTT fleet operation guidelines