01.Design Challenge
How might we efficiently consume and manage food to minimise wastage on a household level?Initial design question
ForgettingFood sits unnoticed until it expires.
Not knowing what to cookWith the ingredients we already have.
No easy trackingLacking an easy way to track inventory and plan meals.
Our team set out to tackle this problem by leveraging AI to create a seamless and helpful kitchen assistant.
02.Ideation & User Flow
I led the team in brainstorming the app’s core functions. We prioritised features that directly addressed the user’s pain points, then mapped these ideas into a comprehensive user-flow diagram to ensure a logical and intuitive navigation structure, with a Smart AI Core at its centre.
User & inputVoice and manual
→Smart AI CoreExpiry and health checks
→Recipe & cooking journeyHealth checks · smart tips · shopping list
→AI planning & insightsRecommend recipes · adjust portions · confirm cooking
01Open & sign inAuthenticate, then join or create a family and fill a health profile.
02Add itemsBy voice or manual input: count, weight or package; fridge, freezer or pantry.
03Smart AI CoreChecks expiry and health needs, recommends recipes, suggests a shopping list, gives tips and reminders.
04CookSelect a recipe, adjust portions or ingredients, confirm cooking; favourites feed back into the loop.
03.Prototyping & Architecture
With the flow defined, I moved into Figma to design the key screens. The focus was on a clean, scannable and encouraging interface.
1 · DashboardGives the user an at-a-glance view of their fridge, upcoming expirations and recipe ideas.
2 · Add itemStreamlined to be as fast as possible, using voice or text.
3 · AI assistantDesigned as a conversational modal, making it feel like a helpful assistant rather than a complex tool.
Dashboard: fridge, freezer and pantry at a glance.
Food inventory interaction
- Edit item
- Unit of measurement (count, weight in g or kg, package), quantity, storage location (fridge, freezer, pantry) and expiry date.
- Ingredient overview
- Each ingredient is tracked in batches with purchase date, expiry date, days left and quantity; one batch can be marked “consume first”.
AI services
- OpenAI API. Natural-language processing (parsing “add item” commands) and recipe generation, fed with inventory, preferences and the blacklist.
- Azure Speech Services. Voice-to-text for natural-language input.
Technical architecture
Client
React NativeOpenAI API
Azure Speech
Server
Node.js
ExpressDatabase
Azure Cosmos DB
Data structure
Cosmos DB holds three collections:
Userid, email, name, languagedietPreferences, dislikedFoodsautoRecommend, recommendationFrequency
UserFavoriteRecipeuserId, recipeIdcreatedAt
SelectedIngredientsuserId, groupIdselectedAt