FoodWise

Home Page
Product Analysis
AI Nutrition Coach
Profile Page
Scan History Page
Login/Starting Page

Inspiration

Grocery shopping shouldn’t feel like a medical exam, yet for one-third of shoppers with diabetes, hypertension, or kidney issues, every label is a stress test. Tiny print, confusing jargon, and misleading “healthy” claims make it hard to stay on track. When multimodal LLMs like Gemini appeared, we asked: what if your phone camera became a pocket dietitian? That idea became FoodWise.

What it does

FoodWise is an intelligent nutrition analysis app designed to empower users, particularly those managing chronic health conditions, to make informed food choices. The app allows users to scan barcodes or nutrition labels for instant food analysis and uses Google Gemini AI to evaluate products based on individual profiles, considering factors like age, weight, height, and medical conditions such as celiac disease or diabetes. It provides a NutriScore assessment with letter grades (A–E) to indicate nutritional quality and delivers a detailed health impact analysis that explains why a product may be suitable or risky for a specific user. FoodWise also includes an AI-powered nutrition chatbot for dietary guidance, tracks scan history with an insights dashboard, and gamifies healthy eating through an achievement system with badges and streak tracking.

How we built it

The frontend was built natively for iOS using SwiftUI and an MVVM architecture to maintain a clean separation of concerns. Custom UI components were developed, including animated backgrounds, modern card layouts, and responsive design elements. A sophisticated tab-based navigation system was implemented, featuring gesture support and smooth animations.

On the backend, the Firebase ecosystem was used for authentication, user profiles, and data storage with Firestore. The app integrates with the Open Food Facts API to access a comprehensive product database and leverages Google Gemini 2.0 Flash for advanced AI-driven analysis, OCR capabilities, and conversational chat. AVFoundation powers real-time barcode scanning and camera functionality.

Key technical features include real-time camera-based scanning, advanced prompt engineering for personalized health analysis, post-processing algorithms for consistent and user-friendly language, an animated splash screen with optimized loading states, robust error handling, offline capability support, and modern iOS design patterns with a custom AppColors theming system.

Challenges we ran into

One of the early challenges was achieving consistency in AI responses, as Gemini AI would alternate between first persona and third person when referencing user health data. This was resolved through refined prompt engineering and text cleanup during post-processing. Performance optimization was another hurdle; initial loading animations caused lag, requiring a redesign with lightweight, efficient animations.

Integrating data from Open Food Facts with AI analysis proved complex, particularly when dealing with missing nutritional data or unrecognized products. Creating a smooth and intuitive user experience while transitioning between scanning, analysis, chat, and history tracking was challenging, as was maintaining a consistent visual design. Additionally, managing complex user profiles in Firebase while ensuring secure, real-time synchronization across devices required careful architectural planning.

Accomplishments we’re proud of

We successfully developed a personalized health technology system that delivers tailored nutrition advice based on individual medical conditions rather than generic recommendations. The app offers a polished, seamless user experience with smooth animations, intuitive navigation, and a professional design comparable to leading commercial health apps.

Our advanced AI integration features sophisticated prompt engineering that ensures consistent, medically relevant analysis with approachable, user-friendly language. FoodWise also offers a comprehensive feature set, combining scanning, analysis, chat, history tracking, achievements, and user profiles into a unified ecosystem. From a technical standpoint, we achieved clean, maintainable code architecture with reusable components and scalable design patterns.

What we learned

Developing FoodWise highlighted the critical importance of precise AI prompt engineering and post-processing to ensure reliable outputs in healthcare applications. We gained valuable insights into designing mobile health interfaces that make complex nutritional information accessible and actionable for users with varying levels of technical expertise.

We learned to handle real-time data processing with multiple API integrations while preserving app responsiveness and developed a deeper understanding of the accuracy, privacy, and trust considerations unique to healthcare applications. Additionally, we advanced our skills in SwiftUI, mastering animations, custom components, and state management for large-scale iOS applications.

What’s next for FoodWise

Future plans include enhancing AI capabilities to provide more sophisticated nutritional analysis, personalized meal planning suggestions, and recipe recommendations tailored to user health goals. We aim to integrate with wearables and Apple Health to deliver holistic insights that combine nutrition with physical activity data.

Social features will be introduced to foster community engagement, allowing users to share healthy product discoveries and connect with others managing similar health conditions. The product database will be expanded to include restaurant menu analysis and support for homemade meal logging.

We also plan to enable healthcare provider integration, allowing medical professionals to monitor and guide patient nutrition through the app. On-device machine learning models will be developed for faster product recognition and improved personalized recommendations. Finally, we aim to support multiple languages and regional nutritional guidelines to serve international markets.