GrocerGPT

Inspiration

Based on the theme of HACKKU, which is about Health & Well-being, we decided to make something that could actually be useful for people

What it does

Grocery Analyzer transforms your shopping receipts into personalized nutrition insights and recipe recommendations. By scanning and analyzing your grocery purchases, the app identifies patterns in your food choices, suggests recipes using ingredients you already have, and provides tailored health advice based on your body measurements and dietary goals. The AI-powered platform helps you make better use of your groceries, reduce food waste, and improve your eating habits without the need for manual tracking or nutritional expertise. Whether you're trying to eat healthier, save money, or just figure out what to cook with what's already in your pantry, GrocerGPT turns your shopping data into actionable food intelligence.

How we built it

GrocerGPT is built on a modern tech stack:

Frontend: Next.js with TypeScript and Tailwind CSS for a responsive, accessible interface

Backend: Node.js with Express for API endpoints

Database: MongoDB for flexible data storage of users, purchases, and recommendations

AI Integration: Google's Gemini API for recipe generation and nutritional analysis

Authentication: NextAuth.js for secure user management

Deployment: Vercel for continuous deployment and scalability

Challenges we ran into

Receipt Data Standardization: Every store formats receipts differently! We spent hours building a flexible parser that could handle variations in formats.

Health Recommendations: Balancing helpful advice without crossing into medical territory was tricky.

API Limitations: The original plan to use specialized nutrition APIs hit a wall when we found most had severe request limits or high costs, we ended up taking the hybrid approach using Gemini's AI with nutritional databases.

What we learned

Quy learned more about Next.js architecture and MongoDB integration, dove deep into AI integration, working with Gemini's API to create personalized recipe and health recommendations

Dan learned more about web development, APIs, database integration and multiple different frameworks

Kaiden learned more about web development, APIs, database integration and multiple different frameworks

Blake tackled the data processing challenge, creating algorithms to extract meaningful patterns from grocery receipts, working with Gemini's API to create personalized recipe and health recommendations, learned more about Next.js architecture and MongoDB integration