ChocoDelivery
Real-time delivery, AI recommendations, autonomous support — for a single-city chocolatier scaling fast.
Started as a 15-day digital menu. Evolved into a production delivery platform with order tracking, driver assignment, fleet management, an AI recommendation engine, and an autonomous support chatbot handling real customer tickets.
- NestJS
- TypeScript
- Go
- Next.js
- Prisma
- PostgreSQL/PostGIS
- Redis
- OpenAI API
- Docker
- Prometheus
- Grafana
From menu to operating system
The original scope: display a menu and save an order. The actual product today: a real-time geo-aware logistics system with live driver tracking, route assignment, and fleet-level KPIs — powered by PostgreSQL + PostGIS and Redis for hot state.
Personalized recommendations
An AI recommendation engine surfaces product suggestions from each user's behavior and order history. No third-party SaaS, no tracking pixels — the signal lives in the same database as the orders.
Autonomous customer support
A chatbot agent diagnoses order issues, reports real-time status, and handles the long tail of repetitive support tickets. Human agents only see escalations that actually need a human.
Observable production
Prometheus + Grafana dashboards for order latency, delivery SLA, and agent resolution rate. Every user-visible failure has a named metric behind it.