Smart Port Logistics APCS

Smart Port Logistics APCS is a Truck Booking Management System integrated with an Intelligent Logistics Assistant to help manage and reduce congestion in the ports by using AI overview

Comment

Inspiration

Global ports suffer from manual booking processes, congestion, miscommunication between actors, and poor real-time visibility. We were inspired by real smart-port initiatives and the question:

What if port logistics systems could understand requests like a human operator — but act instantly, consistently, and at scale?

This led us to combine Smart Port logistics with an AI multi-agent system capable of interpreting natural language and automating booking workflows.

What it does

Smart Port Logistics APCS is an AI-powered booking platform that manages truck gate access in ports.

It allows:

  • Carriers and operators to create, manage, and validate bookings
  • Intelligent handling of time slots, gates, trucks, and capacity

  • Natural-language interaction with an AI system to:

    • create a booking,
    • check availability,
    • modify or cancel reservations,
    • receive alternative slot suggestions automatically

  • Real-time updates via WebSockets for operators

  • Secure authentication using JWT HttpOnly cookies

The system enforces real business rules (capacity limits, booking lifecycle, roles) while remaining flexible and automated.

How we built it

We designed the platform with a clear separation of concerns:

Backend (Core System)

  • NestJS with Prisma ORM
  • PostgreSQL database
  • Strict domain models (Bookings, Gates, TimeSlots, Trucks, Users)
  • Role-based access control
  • Unified booking lifecycle logic
  • REST API + WebSockets (Socket.io)

AI System (APCS)

  • Multi-agent architecture:

    • Orchestrator Agent
    • Intent Classification Agent
    • Entity Extraction Agent
    • Booking Agent
    • Availability Agent

  • Built using Python + FastAPI

  • The orchestrator interprets user intent and delegates tasks to agents

  • Agents interact with the backend API, not the database directly

DevOps

  • Fully containerized using Docker & Docker Compose
  • Easy local deployment for hackathon demos

Challenges we ran into

  • Aligning AI output with strict backend contracts (UUIDs, enums, capacity rules, booking states)
  • Designing an AI system that does not hallucinate actions, but executes validated backend operations
  • Synchronizing real-time events with booking state changes
  • Making authentication secure while still hackathon-friendly
  • Time pressure: building both a production-grade backend and an AI system in parallel

Accomplishments that we're proud of

  • A real, enforceable booking lifecycle (not a mock demo)
  • A working multi-agent AI architecture, not a single chatbot
  • Clean separation between AI reasoning and business logic
  • Real-time notifications for operators
  • A system that could realistically scale to real port operations
  • Full Dockerized setup ready for deployment

What we learned

  • AI systems are far more reliable when treated as orchestrators, not decision dictators
  • Strong backend contracts dramatically reduce AI errors
  • Multi-agent systems outperform monolithic prompts for complex workflows
  • Real-world logistics problems require state, constraints, and validation
  • Designing for hackathons still benefits from production-grade thinking

What's next for Smart Port Logistics APCS

  • Persistent AI memory per carrier/operator
  • Predictive congestion and load forecasting
  • QR / IoT gate integration with real hardware
  • Advanced analytics dashboards
  • Conversational frontend UI
  • Fine-grained audit logs and monitoring
  • Deployment in a real smart-port pilot environment

Built With

Share this project:

Updates