🥗 CulinaMind AI — An Intelligent Kitchen Companion
🌍 Inspiration


Globally, nearly 1.3 billion tonnes of food are wasted every year — and a significant portion of that loss begins at home. Groceries get buried in the fridge, ingredients expire unnoticed, and daily meal decisions become repetitive and exhausting.
CulinaMind AI was built to solve both problems simultaneously. The vision was simple yet powerful:
An AI-powered kitchen companion that understands your pantry in real time, prevents food waste, simplifies meal decisions, and makes cooking effortless and enjoyable.
🚀 What CulinaMind AI Does
CulinaMind AI transforms your kitchen into an intelligent ecosystem.
🧊 Smart Pantry Management
- Track food items with categories, quantities, and expiry alerts.
- Detect surplus and items nearing expiration.
- Add items naturally via AI chat (e.g., “I bought 2kg rice, 6 eggs, and spinach”).
- Structured parsing converts natural language into categorized pantry entries instantly.
🎥 Video → Ingredient & Instruction Extraction
- Upload a cooking video.
- AI extracts ingredients, quantities, and step-by-step instructions.
- Converts unstructured content into actionable recipes.
🤖 Context-Aware AI Assistant (Powered by Google Gemini)
- Real-time pantry-aware recipe suggestions.
- Prioritizes ingredients expiring soon.
- Provides cooking guidance and nutrition insights.
- Fully personalized responses — not generic outputs.
🍽️ AI Meal Planning
- Weekly meal plans based on your pantry inventory.
- Dietary preference customization.
- Waste-reduction optimized planning.
📊 Nutrition Intelligence
- Daily, weekly, and monthly intake tracking.
- Visual nutritional analytics.
- Ingredient-level nutrition breakdown per recipe.
🛒 Smart Cart
- Automatically generates shopping lists.
- Identifies low-stock and frequently used items.
🎙️ Voice-Enabled Cooking
- Hands-free cooking mode.
- Speech-to-text powered by Gemini audio capabilities.
- Seamless step navigation while cooking.
📥 Recipe Import
- Import from URLs or raw text.
- Save to a structured personal cookbook.
👑 Pro Subscription
- Premium features powered by RevenueCat.
- Monthly, Yearly, and Lifetime plans via custom paywall.
🏗️ How We Built It
Frontend
- React Native (Expo SDK 52)
- TypeScript
- React Navigation v6 (Bottom Tabs + Native Stack)
AI Engine
- Gemini 3 (
gemini-3-flash-preview) via@google/genaiSDK Unified AI service powering:
- Chat
- Pantry parsing
- Recipe generation
- Nutrition analysis
- Image generation
- Audio transcription
State Management
- Zustand v4
- 10+ dedicated modular stores (pantry, chat, nutrition, cart, subscriptions, etc.)
UI/UX
- Dark-mode-first design system
- Warm orange primary palette (#F97316)
- Reanimated animations
- Haptic feedback
- Safe-area-aware layouts
Architecture
- Clean service layer for AI operations
- Config layer for environment separation
- Typed navigation system
- Reusable component-driven design
⚙️ Challenges & Solutions
🔧 RevenueCat + Expo Go Conflicts
Native paywall UI was incompatible with Expo Go.
Solution: Built a fully custom PaywallScreen and removed native UI dependency.
🧠 Dual AI Context Problem
Two AI systems operated independently without pantry awareness.
Solution: Redesigned chatWithGemini() to inject dynamic pantry context into every AI call.
🎛️ UI Overlap (FAB Conflict)
Global chat and pantry assistant buttons overlapped. Solution: Repositioned to opposite screen edges.
🧾 Structured JSON Parsing
Ensuring Gemini consistently returned structured JSON required refined prompt engineering and fallback parsing mechanisms.
🏗️ EAS Build Failures
macOS duplicate native files corrupted builds.
Solution: Cleaned native directories and optimized .gitignore.
🏆 Key Accomplishments
🧠 True Context-Aware AI
Every AI call dynamically includes real pantry inventory with expiry prioritization.
🎨 Production-Ready UI
- Cohesive 7-tab navigation
- Smooth animations
- Native feel and responsiveness
⚡ Natural Language Pantry Entry
Conversational item logging converted into structured data with units, categories, and estimated expiry.
🔄 Unified AI Service Layer
Six AI-powered features managed through one consistent architecture.
📚 What We Learned
- Prompt Engineering is Foundational — Structure and context dramatically influence output quality.
- Expo Go Has Native Limitations — Custom builds provide more flexibility.
- State Architecture Must Scale Early — Clear separation simplifies feature integration.
- AI + Structured Data Unlocks Power — Returning JSON alongside natural language enables dynamic UI automation.
🔮 What’s Next
🔔 Smart Expiry Push Notifications
Example: “Your spinach expires tomorrow — here are 3 quick recipes.”
📸 Camera-Based Pantry Scanning
AI-powered grocery detection using Gemini Vision.
🌍 Community Surplus Sharing
Enable neighbors to share excess food and reduce waste locally.
🗄️ Persistent Storage
AsyncStorage / MMKV integration for long-term data retention.
🏪 Grocery Store Integration
Price comparison and one-tap ordering from nearby stores.
📱 Production Launch
Deploy to App Store & Google Play using EAS Build with full subscription integration.
🎯 Vision
CulinaMind AI is not just a meal planner — it is a preventive food intelligence system designed to reduce waste, optimize nutrition, and redefine everyday cooking with AI precision.
It transforms kitchens from reactive spaces into proactive, intelligent environments.
Built With
- app-store-&-google-play-(deployment-target)
- custom
- dark
- design
- eas-build
- eas-build-(production-builds)
- expo-go-(development)
- expo.io
- google-cloud-(gemini-api)
- google-gemini-3-(gemini-3-flash-preview)
- haptic-feedback-apis
- image-generation-(gemini)
- ios-&-android-(mobile)
- local-storage-(asyncstorage-/-mmkv-planned)
- react-native-(expo-sdk-52)
- react-navigation-v6-javascript
- reanimated
- revenuecat-cloud
- revenuecat-sdk
- safe-area-context
- speech-to-text-(gemini-audio)
- theme
- typescript
- zustand-v4
Log in or sign up for Devpost to join the conversation.