PayPilot AI

Dashboard
Compare
AI Search Results
Orders & Spent
Shopping Cart
Product Catalog

Inspiration : Online shopping often means jumping between product pages, comparing specifications, checking prices, and trying to stay within budget. We wanted to make that process easier with an AI assistant that helps shoppers make informed decisions while keeping them in control of every purchase. What it does : PayPilot AI turns a shopper’s request into product requirements, filters and ranks matching products, and explains the trade-offs behind its recommendations. Shoppers can compare options, review a cart, and proceed to PayPal checkout with their approval. How we built it : We designed PayPilot AI around a React and TypeScript frontend, a FastAPI backend, and PostgreSQL for product, order, payment, and audit data. The AI agent interprets shopping requests and uses an allowlisted set of tools. Deterministic application logic handles hard product constraints, ranking, and budget calculations. PayPal checkout is handled through the backend, where payment status can be verified against PayPal’s response. Challenges we ran into : A key challenge was making recommendations useful and trustworthy when product prices and availability can change. We addressed this by grounding recommendations in product data and evidence, and by keeping filtering, ranking, and budget calculations in application logic. We also designed the checkout flow so the shopper explicitly approves a purchase and the backend verifies payment status. Accomplishments that we're proud of : We brought product discovery, comparison, budget awareness, and a shopper-approved payment flow into one coherent project concept. We’re proud of the focus on clear explanations and user control: the agent can recommend and prepare, while the shopper decides whether to buy. What we learned : We learned that a shopping agent needs more than a capable language model. It also needs reliable product data, predictable decision logic, clear explanations, and careful payment-state handling. Trust depends on showing the basis for recommendations and treating payment confirmation as a backend responsibility. What's next for PayPilot AI : Next, we plan to connect verified product sources, build and test the end-to-end shopping experience, and integrate PayPal Sandbox checkout. We also want to evaluate recommendation quality, price freshness, and how effectively shoppers can compare options and complete checkout.