what2eat

Inspiration: $$ \ $$ We've all been there: staring at the fridge after a long day, too tired to cook, and even more tired of scrolling through endless recipes. For busy students and professionals, meal planning often falls apart, leading to unhealthy food choices, repetitive, boring meals, or another night of takeout. That's why we built what2eat: your all-in-one dashboard to take away the decision fatigue and make meal prep actually easy. $$ \ $$ What it does: $$ \ $$ With what2eat, just tell us your nutrition goals and cravings and let us figure out the logistics. In under 60 seconds, it will create a personalized 7-day meal plan that fits your macro counts, dietary restrictions, kitchen equipment, available cook time, and favorite cuisines. You can visualize and track your daily and weekly meals on the dashboard and dive into each meal card to get ingredients, steps, and dietary info. Don't like something? Swap it out by choosing from our database of recipes or ask the AI nutrition consultant for a nutritionally equivalent meal. And the best part? With one tap, your entire grocery list is built and sorted with smart features, substitution suggestions, and more. Think of what2eat as your smart meal-prep partner that connects what you want and what you need. $$ \ $$ How we built it: $$ \ $$ On the frontend, we used React/Next.js to design the UI/UX, including the onboarding wizard for profile set up and selecting dietary preferences and goals, the main dashboard with meal plan grids and expandable elements for viewing recipe details, and the navigation bar for the user profile, and settings. On the backend, we built and used a Python webscraper to scan for public recipes while crediting them, GPT-4o-mini to generate personalized plans from preferences and swap if needed, Supabase for storing user profiles and recipes, and normalized recipe units and consolidated ingredients to the cart. $$ \ $$ Challenges we ran into: $$ \ $$ We faced several challenges including generating recipes that both fit user preferences and take inspiration from scraped recipes, formatting the webscraped recipes into data structure that's readable for the React app, reformatting single React script into structured files for streamlining code management and debugging, and integrating all the different components together. $$ \ $$ Accomplishments that we're proud of: $$ \ $$ We created a pretty but straightforward UI/UX that is simple and intuitive to use and successfully built a system for generating a weekly meal plan unique to each user's needs. $$ \ $$ What we learned: $$ \ $$ We learned about vibe coding with Claude and Cursor through prompt curation and troubleshooting, Git configuration and organization, and database storage and organization with Supabase. $$ \ $$ What's next for what2eat: $$ \ $$ This is just a demo with basic functionalities. We can improve by adding a "cook once, eat twice" mode for leftovers, smart shopping cart that auto adds and sorts based on meal plan, calendar integration with prep reminders, and an AI nutritionist to consult with for fitness goals and nutrition goals, and adjust future recommendations based on meal feedback.