Inspiration
Personal care shopping is overwhelming. There are thousands of skincare, hair care, and beauty products, and most shops make you scroll and filter until you give up. I love beauty and fashion, and I wanted a shop that works the way you'd talk to a friend: "I need a sunscreen under $15" and it just shows you good options. When I saw the PayPal AI hackathon, I knew I wanted to build a complete shopping journey, from asking for a product to paying for it.
What it does
Glow & Care is a personal care and beauty shop with an AI shopping assistant. Shoppers type what they need, like "skincare essentials under $30," and the assistant recommends matching products from a catalog of 31 best-selling items across skincare, hair care, cosmetics, underarm care, sunscreen, and hygiene.
Every product has a photo, size, and price. Shoppers can filter by category, search, build a shopping bag, and check out with PayPal using the PayPal Developer Sandbox, so no real money moves. Orders are saved and can be viewed on a password-protected admin page.
How we built it
- Backend: Node.js and Express
- Frontend: HTML, CSS, and JavaScript, designed in a clean black-and-white style
- Payments: PayPal Orders API (create order, then capture payment) in the sandbox
- AI: Anthropic's Claude API for product recommendations, with a keyword-matching fallback if the AI is unavailable
- Hosting: Render for the live app, GitHub for the code
Challenges we ran into
- Product photos: Many product images from other websites failed to load, leaving blank placeholders. I collected the missing photos myself and added them to the project so they always show.
- Prices from different countries: Some prices came from India and Australia, so I converted them to estimated USD and labelled them clearly.
- Deployment: I first tried GitHub Pages and only the README showed, because the shop needs a running Node server. Moving to Render fixed it.
- Protecting secrets: I accidentally uploaded my
.envfile to a public repo, then deleted it and replaced all my keys.
Accomplishments that we're proud of
- A full flow that works from start to finish: ask the AI, add to the bag, pay with PayPal, and see the saved order.
- A shop that looks polished and works on phones and computers.
- Shipping my first hackathon project of this kind.
What we learned
- How the PayPal Orders API works, including sandbox accounts, creating orders, and capturing payments.
- How to connect an AI model to a real catalog and keep it useful with a fallback.
- Never commit API keys: keep them in
.envand in the host's environment settings. - The difference between hosting static pages and hosting a server app.
What's next for GLOW & CARE
- Real inventory and stock levels
- User accounts and order history
- Prices in local currencies, starting with South Africa and India
- Smarter recommendations based on skin type and past orders
Built With
- anthropic
- claude
- css3
- express.js
- github
- html5
- javascript
- node.js
- paypal
- paypal-sandbox
- render