Commerce Radar

Commerce Radar uses AI and currency trends to help shoppers decide what to buy and where to shop—at home or abroad—with explainable recommendations and a PayPal Sandbox checkout demo.

Comment

Inspiration

Most shopping assistants focus on what to buy. But where you shop matters, too.

Currency movements can change the purchasing power of your money across international markets. Yet consumers rarely consider exchange rates when discovering products, and shopping assistants often lack the evidence to recommend one market over another.

I built Commerce Radar to connect personal shopping interests with global currency intelligence—and help consumers decide not only what to buy, but where to shop.

What it does

Commerce Radar is an AI-powered shopping strategy advisor for US consumers.

It combines purchase history, AI-powered interest analysis, historical exchange rates, and real product listings to recommend one of three strategies:

  • Explore Overseas — Discover relevant foreign-currency products when meaningful FX opportunities are supported by evidence.
  • US Stores First — Prioritize US shopping sources when foreign exchange conditions offer no compelling advantage.
  • Compare Both — Explore domestic and international options when the evidence is mixed or insufficient.

An interactive world map visualizes currency trends and shopping opportunities. Recommended products include observed prices, listing currencies, and transparent explanations. An AI Concierge answers questions and helps users refine their shopping strategy.

Commerce Radar also integrates PayPal Sandbox checkout for clearly labeled fictional demo-merchant products, demonstrating order creation, buyer approval, payment capture, and transaction verification.

How we built it

The frontend uses React, TypeScript, and Vite, with interactive maps and currency charts.

The backend uses Python, FastAPI, SQLAlchemy, and SQLite.

Key integrations include:

  • OpenAI API for purchase-interest analysis, conversational assistance, and personalized explanations.
  • Frankfurter API for daily exchange rates and historical currency comparisons.
  • Channel3 API for structured product discovery, providing observed listing prices, currencies, and merchant sources to support evidence-based shopping recommendations.
  • Web search and curated official product listings for additional product discovery and source evidence.
  • PayPal Orders API (Sandbox) for order creation, buyer approval, and verified payment capture.

A deterministic Python Shopping Strategy Engine combines interest relevance, FX movements, and product evidence to select a recommended shopping strategy.

The AI explains the recommendation, while Python handles numerical calculations and decision rules.

Challenges we ran into

The biggest challenge was connecting currency movements to meaningful shopping recommendations.

A stronger dollar does not automatically mean an overseas product is cheaper than a domestic alternative. Product currencies, seller locations, delivery costs, and historical prices are separate pieces of evidence.

We addressed this by distinguishing verified observations from unknown information and avoiding unsupported savings claims.

We also implemented caching, fallback behavior, and safeguards for external product links, while keeping real product discovery separate from fictional Sandbox checkout.

Accomplishments that we're proud of

  • Built an explainable strategy engine that can recommend domestic shopping—not just overseas products.
  • Combined personalized AI analysis with actual historical FX observations and product evidence.
  • Created an interactive global opportunity map and AI Concierge.
  • Implemented a working PayPal Sandbox checkout with verified completed captures.
  • Demonstrated payment activity in both Sandbox buyer and merchant accounts.
  • Added automated tests for recommendation logic, evidence handling, and payment workflows.

What we learned

AI-powered commerce becomes more trustworthy when language models and deterministic systems work together.

AI is valuable for understanding preferences and explaining recommendations. Numerical calculations, evidence validation, and payment verification require auditable logic.

We also learned that honestly showing uncertainty can be more useful than claiming an unverified bargain.

What's next for Commerce Radar

Future improvements could include verified international shipping and tax estimates, broader retailer coverage, more home currencies, and stronger comparisons of final purchase costs.

Our long-term vision is a personal AI shopping advisor that understands what you love and helps you discover where your money can go further—without sacrificing transparency.

Know What to Buy. Know Where to Buy.

Built With

Share this project:

Updates

Submission history