Project Overview
FletPaq is an end-to-end logistics platform built for the cement industry — designed specifically for Grupo GCC and Cementos de Chihuahua. It manages 60+ fleet units (tractor-trailers, concrete mixers, cement pumps, flatbeds) and predicts equipment failures before they happen using AI models trained on real industrial datasets.
Predictive Maintenance with AI
The heart of the platform. XGBoost models trained on Microsoft Azure PM and Scania Trucks datasets analyze 9 components per unit and output a 0-100 risk score with a color-coded traffic light:
| Component | What it predicts | |---|---| | Brakes | Accelerated wear from Sierra de Chihuahua mountain routes | | Tires | Degradation from desert heat + heavy loads | | Suspension | Stress from bulk cement on rough terrain | | Engine | Progressive wear by mileage and age | | Transmission | Fatigue from constant load + mountain driving | | Electrical | Probabilistic age-based failures | | Hydraulics ★ | Mixer drum system | | Drum Blades ★ | Internal wear from intensive concrete use | | Discharge Chute ★ | Degradation of emptying system |
★ Components exclusive to concrete mixers — the model auto-detects them by unit type.
An embedded AI assistant answers questions about fleet status in natural language ("which units need urgent service?", "what's the state of the mixers?", "recommend a weekly maintenance plan").
Features
- Fleet management — 60+ tractor-trailers, mixers, pumps, flatbeds, trailers
- Load logistics — 90+ real routes across Chihuahua (Cd. Juárez, Delicias, Cuauhtémoc, Parral, Creel)
- GPS tracking — Real-time unit location with active fault codes
- Maintenance — Full history + AI-driven failure prediction
- Billing & collections — Invoice and payment tracking
- Operator PWA — Mobile app for drivers in the field
- Client portal — Clients track their deliveries live
- Role-based access — Admin, operator, client views with different permissions
Tech Stack
- Frontend: Next.js 16, React 19, Tailwind CSS, Framer Motion, Recharts
- Backend: Next.js API routes + Mongoose
- Database: MongoDB Atlas
- AI: Python XGBoost models + Anthropic SDK for the natural-language assistant
- Auth: JWT + bcrypt
- Deployment: Vercel
Demo
Try it live: test / 8896 at fletpaq.vercel.app