Miranda Closet

AI-powered wardrobe and couple styling for real weather, real occasions, and a more sustainable closet. Your smart wardrobe for life, love, and sustainable style.

Comment

About the Project

💡 Inspiration

It started with a simple, universal frustration: standing in front of a closet full of clothes, feeling like you have nothing to wear. But as I talked to friends and couples, I realized the problem was bigger than indecision.

  • People waste 15–20 minutes every morning deciding what to wear.
  • 80% of our wardrobes go unworn.
  • Couples argue because one partner takes too long, or they leave the house looking like they're attending two different events.
  • Fast fashion keeps pumping out cheap clothes while millions of usable garments end up in landfills.

I wanted to build something that didn't just suggest outfits—but understood the real world: the weather outside, the occasion on your calendar, and the harmony between two people. That's how MIRANDA was born.

The name is a playful nod to The Devil Wears Prada—a reminder that style is personal, but it should also be intelligent, sustainable, and a little bit magical.

🧠 What it does

MIRANDA is an AI-powered personal wardrobe ecosystem for individuals and couples. It digitizes your existing clothes, suggests daily outfits based on live weather and occasion, links two wardrobes for coordinated "Couple Harmony" looks, and helps you retire and donate unworn items to local charities. It also generates anonymized, aggregated trend data—the Miranda Index—sold to fashion brands for real-time market intelligence.

🛠️ How we built it

MIRANDA is a native Android app built with Kotlin and Jetpack Compose, generated and iterated using Google AI Studio's native Android vibe coding capability.

Core Stack

Layer Technology
Frontend Kotlin, Jetpack Compose, Material 3
Backend Appwrite (Auth, Database, Storage, Messaging)
AI Engine OpenRouter (Gemini / GPT-4o-mini) for outfit generation
Weather OpenWeatherMap API
Monetization RevenueCat (subscriptions, paywalls, entitlements)
Notifications OneSignal (engagement campaigns)
CI/CD Codemagic (builds and deployment)

Key Features Built

  • Digital Wardrobe: Users snap photos of clothing; the app tags type, color, and formality.
  • Weather-Intelligent Daily Stylist: Outfit suggestions based on live temperature, rain, wind, and humidity.
  • Occasion Modes: Work, Party, Date, Gym, Chill, and Custom.
  • Couple Harmony (Modo Sintonia): Links two wardrobes and generates coordinated looks—including a "conflict alert" when colors or formality clash.
  • Wardrobe Health & Retirement: Flags items unworn for 3+ months and offers donation, swap, or recycle options.
  • One-Click Donation Pipeline: Geolocation finds nearby charities and tracks impact.
  • Miranda Index (B2B): Anonymized, aggregated trend data sold to fashion brands.

The Math Behind the Impact

When a user retires an item through MIRANDA, we estimate the environmental benefit:

[ \text{Textile saved (kg)} = \text{items retired} \times 0.5 \text{ kg} ]

[ \text{CO}_2 \text{ avoided (kg)} = \text{Textile saved} \times 5.0 ]

So retiring just 10 items saves roughly:

[ 10 \times 0.5 = 5 \text{ kg of textile} ] [ 5 \times 5.0 = 25 \text{ kg of CO}_2 ]

That's the equivalent of driving about 100 km less in a standard car.

🧗 Challenges we ran into

  • Prompt Engineering for Outfit Generation: Getting the AI to return useful combinations in a consistent JSON format took dozens of iterations. I had to carefully craft prompts that included wardrobe items, weather, occasion, and—for couples—the partner's wardrobe.
  • Couple Linking Logic: Designing the invite flow, managing two user profiles, and ensuring real-time synchronization between two devices was more complex than expected. Appwrite's real-time capabilities helped, but state management was tricky.
  • Privacy vs. B2B Value: I wanted to build the Miranda Index without compromising user trust. I implemented strict anonymization (aggregating data only across groups of 1,000+ users) and made opt-in transparent.
  • Time Constraints: Building a full ecosystem in 8 weeks meant prioritizing ruthlessly. I focused on the core loop—digitize, generate, wear, retire—and deferred advanced features like virtual try-on and P2P swapping to post-launch.
  • RevenueCat Integration: Setting up three tiers (Basic, Pro, Harmony) with different entitlements and a 7-day free trial required careful configuration. RevenueCat's documentation and the Shipaton perks made it manageable.

🏆 Accomplishments that we're proud of

  • A fully functional Android app that actually works—built from scratch during the hackathon.
  • The Couple Harmony feature, which no other wardrobe app offers.
  • A triple-bottom-line business model: consumer subscriptions, B2B data, and social impact.
  • Integrating six different APIs (Appwrite, RevenueCat, OpenWeatherMap, OpenRouter, OneSignal, Codemagic) into a cohesive product.
  • A clear path to launch on Google Play by the Shipaton deadline.
  • The Miranda Index concept, which turns everyday wardrobe behavior into a valuable B2B trend-forecasting product.

📚 What we learned

  • AI is a styling assistant, not a replacement for taste. The best results came when the AI worked with the user's existing wardrobe, not against it.
  • Weather and occasion are non-negotiable filters. A beautiful outfit is useless if it's wrong for the temperature or the event.
  • Couples are a powerful design constraint. The "Harmony" feature forced me to think about two wardrobes as a single system—color coordination, formality balance, and shared scheduling.
  • Data can be ethical and valuable. The Miranda Index only works if users trust that their data is anonymized and used responsibly.
  • Vibe coding is real. Using Google AI Studio to generate a native Android app from natural language prompts was a game-changer. It let me focus on product logic and user experience instead of boilerplate code.

🚀 What's next for Miranda Closet

  • Virtual Try-On: Use AI image generation to render outfits on the user's body.
  • P2P Swapping Club: Local peer-to-peer clothing exchange with in-app messaging.
  • Miranda Index Dashboard: A real-time B2B portal for fashion brands.
  • Geographic Expansion: Localization for Europe and Asia.
  • Partnerships: Collaborate with sustainable fashion brands and charities.
  • Launch on Google Play: Public release with full subscription tiers via RevenueCat.

MIRANDA is more than an app. It's a movement toward smarter, more sustainable, and more harmonious style.

Dress in harmony—with your life, your love, and your world.

Built With

  • ai
  • android
  • appwrite
  • backend-as-a-service
  • codemagic
  • fashion
  • gemini
  • generative-ai
  • google-ai-studio
  • gpt-4o-mini
  • jetpack-compose
  • kotlin
  • llm
  • machine-learning
  • material-3
  • mobile-app
  • onesignal
  • openrouter
  • openweathermap
  • push-notifications
  • revenuecat
  • stripe
  • subscriptions
  • vibe-coding
  • weather-api
Share this project:

Updates

Submission history