Truly Multimodal
Text, call, dictate, or stream video—just like FaceTime. Hands-free voice mode with live video streaming and an animated chef avatar lets you cook while you chat.
iOS App
Heard, Chef!
Your AI cooking assistant powered by Gemini. Chat like iMessage, cook hands-free with voice mode, and manage your recipes, fridge, and pantry—all on your phone.
Features
Knows Your Kitchen
Maintains a complete record of your available ingredients, references your saved recipes, and remembers your allergies and preferences.
Privacy First
All your recipe, fridge, and pantry data stays on your phone with SwiftData. Only chat messages are sent to Gemini for AI processing.
How It Works
1
Ask about your kitchen
"Do I have enough eggs for carbonara?"
The AI calls get_ingredient and list_ingredients tools to query your SwiftData inventory. You'll see a live status chip: 'Checking Inventory...' followed by 'Found: 6 eggs'. It then searches your recipe book and suggests matching dishes.
2
Cook with voice guidance
"How do I know when these onions are caramelized?"
Switch to hands-free voice mode with live video streaming. The AI streams responses via Gemini's multimodal live API using PCM 16-bit audio at 16kHz. An animated chef avatar reacts to your voice. Stream video of your pan or snap a photo mid-cooking for instant visual analysis.
3
Manage inventory effortlessly
Snap a photo of your grocery receipt
The AI parses items using vision, normalizes quantities ('a bunch of cilantro' → count: 1, category: produce), and calls add_ingredient for each item. All data saves locally to SwiftData—never leaving your phone.
4
Build your recipe collection
"Save this recipe for shakshuka"
The AI calls create_recipe with ingredients and steps, tags it appropriately, and adds missing items to your shopping list. Later, ask 'What can I make tonight?' and it'll call search_recipes and check_recipe_availability to suggest meals based on what you have.
Frequently Asked Questions
What are the requirements?
iOS 17.0+, Xcode 15.0+, and a Google Gemini API key with multimodal live access capabilities.
Is it available on the App Store?
Currently in development. You can build and run it yourself from the GitHub repository. Free and open source under AGPL-3.0 with Commons Clause.
Will local models be supported?
Potentially in the future if Google releases local model capabilities. The architecture is designed to be flexible for different AI providers.
Built With
Swift & SwiftData
Native iOS with local-first data storage
Gemini 2.0 Flash
Real-time multimodal AI streaming
PCM 16-bit Audio
Low-latency voice streaming at 16kHz
Tool Calling
AI manages recipes, fridge, and pantry data
Get Started
Free and open source. Build it yourself or follow along with development.