About PayPilot
Inspiration
Online shopping is convenient, but finding the right product within a budget can still be frustrating. Users often spend too much time comparing products and figuring out which option actually meets their needs.
We wanted to make this process smarter. What if users could simply describe what they need and let AI help them find suitable products?
That idea inspired PayPilot — an AI-powered shopping assistant designed to make product discovery easier while keeping purchasing decisions in the user's hands.
What We Built
PayPilot combines AI-powered product recommendations, shopping validation, and PayPal payment integration in one application.
Users describe their shopping needs in natural language, and Gemini AI helps recommend products from a controlled catalog.
Key features include:
- Catalog validation: Recommendations are restricted to available products.
- Budget validation: Products are checked against the user's budget.
- Relevance checks: Suggestions are evaluated against shopping requirements.
- Product transparency: Product limitations are communicated clearly.
- Human approval: Purchases require explicit user authorization.
- PayPal integration: Users can experience the checkout flow through PayPal Sandbox.
How We Built It
We built PayPilot using Next.js and TypeScript, with Google Gemini powering AI recommendations and PostgreSQL supporting order storage.
Deterministic validation logic helps enforce catalog, budget, and relevance rules. PayPal Developer Sandbox handles order creation and payment capture, allowing us to test the complete checkout experience without real charges.
Challenges We Faced
Our biggest challenge was balancing AI flexibility with strict shopping rules. AI recommendations can sometimes be irrelevant or unpredictable, so we worked on validation checks and fallback handling.
Integrating payment authorization and capture was another challenge. We designed the workflow so that AI recommendations cannot independently authorize a purchase.
Bringing AI, validation, and payments together required careful testing across the entire shopping journey.
What We Learned
We learned that building trustworthy AI applications requires more than integrating an AI model. Structured data, deterministic validation, transparent recommendations, reliable error handling, and human control are equally important.
We also gained practical experience integrating Gemini AI, building full-stack workflows, and implementing PayPal Sandbox payments.
Our Vision
We believe shopping should be smarter, more transparent, and more user-controlled.
With PayPilot, we aim to bring AI-powered product discovery and secure digital payments together.
PayPilot — Smarter Shopping. Safer Decisions.
Built With
- agentic
- agents
- api
- authentication
- commerce
- css
- development
- e-commerce
- gemini
- generative
- intelligence
- javascript
- next.js
- node.js
- payment
- payments
- paypal
- react
- rest
- sandbox
- secure
- shopping
- sqlite
- tailwind
- typescript