Inspiration
Online shopping often requires users to visit multiple websites, compare prices, evaluate sellers, and identify hidden risks before making a purchase. This process can be time-consuming and confusing.
We wanted to explore a better approach: What if users could simply describe what they need and delegate the research to an AI-powered shopping assistant?
This idea inspired GHOSTBUY AI — Don't Shop. Delegate.
What it does
GHOSTBUY AI is an AI-powered shopping assistant that transforms natural-language shopping requests into structured product research and recommendations.
Users describe their shopping mission and set a maximum budget. The system retrieves product listings, compares available information, evaluates potential purchase risks, and classifies results into:
- Eligible
- Needs Verification
- Rejected
Users can review the recommendations and initiate PayPal Sandbox checkout with explicit approval.
How we built it
We developed GHOSTBUY AI using Python and Flask for the backend, HTML, CSS, and JavaScript for the frontend, Google Gemini for AI-powered analysis, and SerpApi Google Shopping for product discovery.
A dedicated risk-analysis engine evaluates available product and seller information. PayPal Sandbox REST APIs provide the checkout integration.
The application follows a modular architecture that separates product research, AI analysis, risk evaluation, and payment processing.
Challenges we faced
One of our biggest challenges was handling incomplete and inconsistent product listings. A low price does not necessarily mean a safe purchase, so we designed separate classifications for eligible, verification-required, and rejected listings.
Integrating PayPal Sandbox also required careful handling of API credentials, order creation, buyer approval, and payment verification.
Another challenge was making AI recommendations transparent rather than presenting every search result as trustworthy.
What we learned
Building GHOSTBUY AI helped us understand how AI agents can interact with external APIs and real-world shopping data.
We gained practical experience with backend development, API integration, product data processing, risk evaluation, payment workflows, and building a complete web application.
What's next
We aim to improve seller verification, product specification matching, real-time availability checks, and payment reconciliation to make autonomous shopping more reliable.
Our vision: Make online shopping smarter, safer, and simpler — while keeping the buyer in control.
