ClawPoints 爪爪積分

Play, Earn, and Redeem - Smarter with AI.

Comment

Inspiration

Working in Hong Kong’s retail industry, I help small businesses with POS solutions. One client runs a consignment shop and recently opened a claw machine arcade. They wanted a member points system to engage customers, but found existing Taiwan solutions is not localized for Hong Kong.

With only three days before the AWS AI Hackathon deadline, I decided to create a prototype using AWS + Kiro to test if I could quickly deliver a localized, AI-powered loyalty platform for small retailers.

What it does

ClawPoints 爪爪積分 is designed as a SaaS membership and loyalty system tailored for claw machine shops and small retailers in Hong Kong.

Customers: Log in with username + password. See current points and QR code. Track history of earned and redeemed points. Optionally chat with an AI chatbot for FAQs.

Shop Staff: Scan item barcodes → add points to customers (for returned clawed items). Scan customer QR codes → add promotional points or deduct points for prize redemption. Shop Owners (Admin Dashboard): Manage branches, prizes, and inventory. View customer history (earnings & redemptions). Enable or disable AI features like chatbot support.

Optional AI Features: Amazon Lex chatbot to answer customer questions. (in demo) Amazon Forecast to predict prize stock shortages. (future - WIP) Amazon Personalize to recommend prizes to customers. (future - WIP)

How we built it

Demo site - Development Tooling: Kiro: AI assistant to bootstrap the project structure, React components, and AWS configuration. ChatGPT: It helps to clarify business process, user requirement and system specification and the use of AWS tools. Amazon Q Developer: Try a while but Kiro is more user-friendly to new users.

(WIP) Production: Frontend: React + Tailwind CSS, hosted on AWS Amplify. Backend: AWS Lambda functions behind API Gateway. Database: Amazon DynamoDB with tenant_id for multi-tenant SaaS design. Authentication: Amazon Cognito (username + password only).

AI: Amazon Lex → AI chatbot. Amazon Forecast → inventory prediction. Amazon Personalize → recommendation engine.

Challenges we ran into

Time pressure: I had less than two days to build a working prototype before the deadline. Learning curve: Kiro, Amazon Q, AWS account tools, API study. It takes time for the registration, installation and the use. Data model design: Separating “items” (earned points from clawed toys) vs. “prizes” (redeemed with points) into different DynamoDB tables. Items are clawed toys returned to staff to add points, while Prizes are rewards members redeem to spend points and reduce stock. Localization: Adapting the solution for Hong Kong (bilingual support, WhatsApp/IG integration).

Accomplishments that we're proud of

Built a prototype in less than two days. Demonstrated a real use case from Hong Kong’s retail industry (claw machine shop). Integrated AWS AI services into a practical loyalty system. Showcased how Kiro can accelerate real product development for small businesses.

What we learned

How to rapidly prototype with Kiro. The importance of data modeling in loyalty systems (separating earning vs. redemption flows). How AI services (Lex, Forecast, Personalize) can bring real value to small retailers. That Kiro + Amazon Q Developer can drastically shorten the development cycle, making Hackathon-scale prototyping realistic.

What's next for ClawPoints 爪爪積分

Production-ready SaaS: Polish the prototype and onboard real Hong Kong merchants. Payment & Billing: Add Stripe integration for shop owners to subscribe to different plans. More AI Insights: Customer segmentation with Amazon SageMaker. Sentiment analysis of customer chatbot conversations. Expand beyond claw machines: Apply to claw machines shops, F&B loyalty programs, and shopping mall campaigns. Localization: Full Cantonese + English chatbot support and WhatsApp integration.

Built With

Share this project:

Updates