FreightIQ: Intelligent Multi-Agent Freight System

Autonomous multi-agent freight quotation engine combining ML dynamic pricing, automated customs compliance, and real-time corridor risk intelligence.

Comment

Inspiration

Global trade moves trillions of dollars of goods annually, yet freight quote generation remains heavily reliant on fragmented spreadsheets, opaque carrier tariffs, and slow manual cross-checks. A single cross-border quotation can take anywhere from 4 to 48 hours as logistics teams manually calculate chargeable weights, assess customs tariffs across thousands of HS codes, verify geopolitical and weather risks, and negotiate carrier spot rates.

We asked ourselves: What if a customer or freight forwarder could receive an auditable, fully compliant, and dynamic market quotation in under 20 seconds?

This inspired us to build FreightIQ (Cargo Genius)—an end-to-end autonomous multi-agent platform that streamlines freight forwarding across Ocean, Air, Road, and Rail corridors into a seamless, intelligent workflow.

How We Built It

FreightIQ is built on a modern decoupled architecture featuring an event-driven React + Vite single-page application and a robust Django REST framework backend orchestrator coordinating 5 specialized autonomous agents.

1. The Autonomous 5-Agent Multi-Verification Pipeline

When an enquiry is submitted, our centralized Quote Orchestrator coordinates five parallel specialized agents through an auditable state machine:

  • Routing & Corridor Agent: Resolves multimodal waypoints across ocean ports, air cargo terminals, road hubs, and inland container depots (ICDs).
  • ML Dynamic Pricing Agent: Employs gradient-boosted decision trees trained on historical spot contracts to predict market-clearing rates alongside deterministic cost-plus policies: $$\text{Quotation Price} = \text{Base Rate} \times (1 + \alpha_{\text{fuel}} + \beta_{\text{congest}}) + \text{Handling Fees} + \text{Margin}$$
  • Weather & Delay Assessment Agent: Evaluates real-time oceanic radar, typhoon tracks, and monsoon indices along the voyage corridor to predict delay probabilities.
  • Customs Compliance & RAG Agent: Matches 6-digit Harmonized System (HS) codes against international regulatory databases using vector embeddings to automatically generate mandatory document checklists (e.g., Phytosanitary certificates, HazMat declarations, Bill of Lading).
  • Composite Risk & Insurance Engine: Computes a transparent $0 - 100$ shipment risk index derived as a weighted summation across operational dimensions: $$\text{Risk}{\text{composite}} = \sum{i=1}^{n} w_i \cdot R_i \quad \text{where} \quad \sum_{i=1}^{n} w_i = 1.0$$

2. Role-Based Collaborative Workspaces

FreightIQ provides dedicated, tailored portals for every stakeholder in the logistics lifecycle:

  • Customer Portal: Step-by-step 4-mode enquiry wizard with interactive weight calculations and 20-second instant quote generation.
  • Freight & Company Agent Portal: Multi-tab operations desk for reviewing incoming requests, verifying risk scores, and generating downloadable PDF quotations.
  • Company Manager Portal: High-level ledger for margin policies, approval rules, and freight team productivity tracking.
  • Customs Officer Portal: Document inspection queue and regulatory alert clearance dashboard.
  • Admin Hub: Master data management for routes, carriers, surcharge policies, and audit trails.

Challenges We Faced

  1. Precision in Pricing & Rounding: Standard floating-point arithmetic can introduce hazardous rounding drift when calculating multimillion-dollar container shipments across multiple currencies. We engineered dedicated Decimal-safe money abstractions and strict ISO-4217 currency conversions.
  2. Harmonizing Multimodal Transport Locations: Structuring unified origin/destination dropdowns across 4 completely different transportation modalities (Airports, Ocean Ports, Road Freight Centers, and Rail Terminals) required designing a compact, ergonomic selection architecture to prevent UI overflow while maintaining clear modality identification.
  3. Low-Latency Agent Orchestration: Executing parallel API evaluations across routing algorithms, ML pricing models, and compliance checks simultaneously without blocking the user interface required optimizing our orchestrator to complete the full 5-agent verification loop in under 20 seconds.

What We Learned

  • Micro-frontend Ergonomics: How to craft interactive, stateful dashboards that preserve clean empty states and progressively reveal complex financial calculations only when valid cargo inputs are provided.
  • Explainable AI in Enterprise Logistics: Pricing algorithms cannot be black boxes. Logistics managers need full transparency into why a surcharge was applied, requiring us to pair ML spot predictions with deterministic rule comparisons.
  • State Machine Reliability: Designing atomic transitions ensuring no quotation can be issued without prior compliance sign-offs, risk assessment, and managerial margin approvals.

What's Next for FreightIQ

  • IoT Telemetry & Satellite GPS Integration: Live tracking container sensors for temperature-controlled reefer cargo and real-time ETA recalculations.
  • Smart Contract Automated Invoicing: Integration with blockchain-based digital Bills of Lading for instantaneous milestone escrow release upon cargo delivery.
  • Autonomous Carrier Bidding Engine: Enabling reverse carrier spot auctions where approved carriers can automatically bid on pending container bookings.

Built With

Share this project:

Updates

Submission history