FreightRate AI
May 2026 - In Progress
A containerized logistics-focused freight quote system designed for cloud deployment.
Overview
FreightRate AI is an in-progress backend-focused logistics SaaS project for calculating and comparing freight quotes across multiple carriers. The project is designed around practical logistics rules: zones, carrier pricing, chargeable weight, minimum charges, surcharges, and taxes.
The goal is not only to return a number. A useful freight quote system should explain why a rate was selected and how the final amount was built. That is where the AI explanation layer fits: the deterministic backend calculates the quote, and the AI layer helps turn the breakdown into clear language.
I am using this project to practice modern backend architecture with .NET 8, PostgreSQL, Docker, Redis caching plans, and deployment thinking around Google Cloud Run.
Problem
Freight pricing is rarely a fixed-rate calculation. A quote can change based on origin and destination, zone mapping, actual weight, volumetric weight, chargeable weight, carrier-specific rules, fuel surcharge, GST, minimum charges, and business-specific exceptions. If those rules are not modeled clearly, the system becomes hard to test and harder to explain to users.
My Role / Contribution
I am designing the backend architecture, quote flow, data model, service boundaries, Docker setup, Redis caching plan, and AI explanation layer. My focus is on keeping the pricing calculation deterministic and testable while using AI only to explain the calculated result.
Key Features
- Multi-carrier quote flow for comparing freight rates.
- Origin, destination, and zone-based pricing support.
- Chargeable weight calculation using actual and volumetric weight.
- Surcharge, minimum charge, fuel, and tax-aware rate composition.
- Carrier pricing rule modeling for different shipment scenarios.
- Dockerized backend setup for consistent local development.
- Redis caching plan for repeated lookup data such as zones and carrier rules.
- AI-assisted explanation layer for making quote outputs easier to understand.
Challenges
- Keeping pricing logic explicit enough to test while still flexible for carrier-specific rules.
- Designing data models that can support changing zones, charges, and date-bound pricing without brittle conditionals.
- Separating calculation correctness from explanation generation so AI does not become the source of truth.
- Planning Redis caching around real lookup patterns instead of adding caching prematurely.
Learnings
- Backend-heavy SaaS products need clear service boundaries around business rules.
- Caching plans should start from real read patterns, not from adding Redis by default.
- AI features work best as explainability layers when deterministic business logic remains authoritative.
- Pricing systems become easier to reason about when each calculation step produces a clear intermediate result.
Architecture / Flow
- A quote request comes into the ASP.NET Core Web API with shipment details such as origin, destination, dimensions, and weight.
- Validation checks required fields and rejects invalid shipment inputs before pricing begins.
- The service layer resolves the shipment zone and selects applicable carrier pricing rules.
- Pricing services calculate volumetric weight, chargeable weight, base charges, surcharges, taxes, and final totals.
- PostgreSQL stores carrier, zone, and pricing configuration, while Redis is planned for hot lookup data that is read often.
- The API returns a structured quote breakdown, and the AI explanation layer turns that breakdown into a plain-language explanation.
- The application is being designed for containerized deployment with Docker and Google Cloud Run.
Tech Stack
Backend
Data & Caching
Deployment & Tooling
Next Improvements
- Add realistic carrier and zone seed data for better testing.
- Build quote history and comparison views.
- Add unit tests around chargeable weight, surcharge, and minimum charge rules.
- Deploy the API to Google Cloud Run once the core quote flow is stable.
Links
No public live demo or repository is available for this project yet.