🍳 Mise: Your AR Cooking Assistant

💡 Inspiration

We have all had that moment: opening the fridge, staring at a random assortment of ingredients, and thinking, "I have nothing to eat." I wanted to eliminate that decision fatigue entirely. I also wanted to solve a problem every home cook faces: trying to scroll through a recipe on a phone while your hands are covered in flour or sauce.

Mise (short for Mise en place) bridges the gap between your physical kitchen and digital guidance. By leveraging the Meta Spatial SDK, I built an assistant that lives inside your kitchen without taking up physical space, making cooking effortless, clean, and genuinely fun.


✨ Key Features & Experience Design

Mise transforms your kitchen into an intelligent, immersive environment.

Feature Description
🤖 AI Sous-Chef (Vision) Have a question? Leveraging the Passthrough Camera API, you can ask our Voice Assistant anything. It can "see" your pot to tell you if the water is boiling or if your potatoes are cooked.
🧠 AI Recipe Disks Don't know what to cook? Simply populate your digital pantry, and our AI Engine generates recipe "disks" based on exactly what you have, your dietary needs, and how much time you have.
🖐️ Hands-Free Cooking No more dirty phone screens. Follow step-by-step spatial instructions with Hand Tracking. Interact with floating timers and tips without touching a thing.
🗑️ Zero Waste Mise tracks your digital inventory and suggests meals to cook before your food expires, saving you money and reducing waste.

"Mise brings your kitchen into the future — giving you a sous-chef that can see, think, and guide you."


🛠️ How We Built It & Technical Implementation

We built Mise as an AR-first experience tailored for the Meta Quest 3/3S. We heavily utilized the Meta Spatial SDK to ensure high performance (60 FPS+) and reliable tracking.

Core Technologies

  • Meta Spatial SDK: Used for spatial anchors (locking UI panels to your walls), hand tracking for hygiene-friendly control, and passthrough environment integration.
  • Passthrough API & Multimodal AI: We pipe visual data from the Quest’s cameras to a Vision LLM, allowing the user to ask context-aware questions about the physical food in front of them.
  • Interactive UI: Custom Android-style components built for Horizon OS that allow for rapid pantry management and intuitive recipe navigation.

🧗 Challenges & Learnings

  • Spatial UI Alignment: Getting VR UI panels to feel naturally placed in a real kitchen without obstructing the user's view of dangerous items (knives/stoves) required careful design and calibration.
  • Latency vs. UX: AI recipe generation and visual analysis can be slow. We optimized caching and "debouncing" AI calls so the user never feels stuck waiting inside the headset.
  • Instruction Translation: Translating flat text recipes into 3D animated guidance was a major UX hurdle. We iterated several times to make the instructions feel native to spatial computing rather than just floating text.

🚀 What's Next for Mise

We are just getting started. Here is our roadmap for the next updates:

  • [ ] Smart Grocery Lists: Auto-generation based on low inventory.
  • [ ] Hardware Sync: Integration with smart fridges and ovens for auto-timers.
  • [ ] Community Cookbook: Share your generated "Recipe Disks" with other users.

Cooking shouldn't be stressful. With Mise, it's the future.

Built With

Share this project:

Updates