ScentWise — AI Fragrance Advisor

AI-powered fragrance recommendations with a database of 65,000+ perfumes, 101 celebrities, and 6 AI recommendation modes.

Features

Free (no purchase needed):

• Database Explorer — search & browse 65,000+ fragrances with notes, accords, ratings, and gender filters
• Celebrity Fragrances — browse 101 celebrities and what they actually wear (with database cross-references)

Premium ($10 one-time — lifetime access):

• AI Chat Advisor — ask about dupes, seasonal picks, budget rotations, blind buys (with database context)
• Style Scan — upload a photo, get fragrance matches for your aesthetic
• Zodiac Match — scents matched to your birth sign's energy
• Music → Fragrance — your playlist reveals your scent identity
• Style Match — streetwear, minimalist, goth? Get matched scents
• 500 AI queries/month

Deploy to Vercel (10 minutes)

Step 1: Get your Gemini API key

1. Go to https://aistudio.google.com/apikey
2. Create a new API key (free)
3. Copy it

Step 2: Push to GitHub

1. Go to github.com → click "+" → New repository → Name it "scentwise"
2. Click "uploading an existing file"
3. Unzip this folder and drag ALL files in (keep the api/ and public/ folders)
4. Click "Commit changes"

IMPORTANT: Make sure the folder structure is at the ROOT of the repo:

scentwise/          ← your repo
├── api/
│   └── recommend.js
├── public/
│   └── index.html
├── package.json
├── vercel.json
└── README.md

NOT nested inside a subfolder.

Step 3: Deploy on Vercel

1. Go to vercel.com/new → Sign in with GitHub
2. Import your "scentwise" repo
3. Before clicking Deploy → expand "Environment Variables"
4. Add: Name = GEMINI_API_KEY, Value = your key
5. Click Deploy

Your site goes live at your-project.vercel.app in ~30 seconds.

Step 4: Set up payments (Lemon Squeezy)

1. Go to lemonsqueezy.com → Create account
2. Create a product: "ScentWise Premium" as a one-time $10 purchase (not a subscription)
3. Copy the checkout URL
4. In public/index.html, find LEMON_URL and replace with your checkout URL
5. Set checkout success redirect to: https://your-site.vercel.app/?paid=true
6. Push the change to GitHub (Vercel auto-redeploys)

Revenue Projection

At $10 one-time with 200 buyers = $2,000 in revenue Gemini API cost: ~$0.001 per query × up to 500 queries × 200 users = ~$100/month at peak usage

Architecture

• Frontend: Single HTML file with 65K fragrances embedded client-side (searches are instant, no API needed)
• Backend: Vercel serverless function calling Google Gemini (model set by GEMINI_MODEL, default gemini-3.5-flash) (only for AI features)
• Payments: Lemon Squeezy handles one-time purchases
• The database explorer and celebrity browsing work completely offline — zero API cost
• AI features send relevant database context to Gemini for accurate recommendations