PER FIT

Log in page
Upload page
Welcome page
Suggestions page
Settings page

PERFIT — Stop staring, start wearing
Your digital closet + AI stylist in one app

Inspiration Every morning, millions of people stand in front of closets full of clothes and say “I have nothing to wear.” I was doing the same thing.

I realized the problem isn’t lack of clothes — it’s lack of visibility and decision fatigue. Meanwhile, 80% of a typical closet goes unworn, and people keep buying duplicates. We wanted to build something that helps you use what you already own, not buy more.

What it does Perfit turns your physical closet into a searchable, smart digital closet and gives you outfit ideas you can actually wear.

Core features: Snap & Organize: Snap photos of your clothes or import from shopping emails. AI auto-categorizes by type, color, season, and tracks wear count so you stop buying the 5th black t-shirt. Outfit Builder: Drag-and-drop canvas to build outfits, or get AI suggestions like “You wore this jacket twice this month. Pair it with these 3 skirts you haven’t worn.” Save outfits to your calendar. Discover Feed: TikTok-style feed of outfits filtered by vibe: Minimal, Streetwear, Office casual, Budget under $50. Every item has an “I own similar” button linked to your closet. AI Try-On: Upload one full-body photo and AI swaps outfits on you in 3 seconds. Built for sharing to Instagram Stories. Stick Features:

Mood Picker: “I feel lazy but cute” → 3 outfits from your closet
Packing List Generator: Plans outfits for trips using only your closet
Price Per Wear: See that $200 boots cost $5 per wear after 40 uses

How we built it MVP stack: Frontend: React Native for iOS/Android, for fast 1-2 tap interactions AI Vision: GPT-4V for photo tagging and clothing categorization Backend: Firebase for auth, storage, and real-time sync AI Recommendations: Python + embeddings for outfit matching based on wear history, weather, calendar AI Try-On: Stable Diffusion + ControlNet for low-latency virtual try-on Weather/Calendar: OpenWeather API, Google Calendar API

We prioritized speed: 90% of actions are 1-2 taps. If it takes 5 taps to make an outfit, people bounce.

Challenges we ran into Clothing detection: Clothes on hangers vs. crumpled on beds confuse models. We solved it by fine-tuning prompts and using background removal. Outfit coherence: AI kept suggesting clashing colors. We added style rules + user feedback loop via swipe-to-like/dislike. Latency for try-on: Full AI try-on was too slow for mobile. We shipped a manual builder first and made try-on async with push notifications.

Accomplishments that we're proud of Built a working closet digitization flow in under 48 hours using GPT-4V Got AI outfit suggestions to feel “scarily accurate” after 10 swipes of training Designed a UI that feels like Depop meets Apple Photos — clean, fast, image-first Created a “Price Per Wear” feature that immediately changed how testers thought about shopping

What we learned People don’t want another marketplace. They want to feel good about what they already own. Speed beats features. A fast manual builder gets used more than a slow “perfect” AI try-on. The sustainability angle hits hard when you show real numbers: “You saved $300 this month by rewearing stuff.”

What's next Phase 2 features: Real-time AR try-on using phone camera Closet swap/resell with friends — see what friends aren’t wearing and swap or sell Offline mode for checking your closet in stores “Style DNA” report after 2 weeks: “You wear 80% neutral colors, love oversized fits”

Goal: Make PERFIT the app you open before Instagram, Amazon, or your closet.

Built for: Anyone tired of decision fatigue and wanting to get more use out of the clothes they already paid for.