-
-
Shopper View: YouCam AI skin diagnostics & Gemini fragrance curation.
-
Boutique Studio: custom YouCam API key management and fragrance catalog control.
-
Real-time facial biomarker analysis (oiliness, moisture, redness) with dynamic AI zone masks.
-
Personalized fragrance curation: matching skin chemistry to perfume concentration and notes.
Inspiration
I adore complex, multi-layered perfumes with intricate note progressions. This past summer, I bought a fragrance that smelled divine on a paper blotter in the boutique. But once I applied it to my own skin, the story completely changed: the delicate notes never fully bloomed, and the scent vanished far quicker than I had hoped.
When I tried consulting AI for fragrance advice, the recommendations were always generic and frustrating — because the AI lacked the single most critical variable: objective biological data about my actual skin. Fragrance molecules are lipophilic, meaning individual epidermal lipids, moisture levels, and pore texture act as the physical anchors for scent evaporation.
Perfect Corp.'s YouCam Skin AI provided the exact missing piece: clinical-grade, non-invasive epidermal biomarkers. By pairing YouCam Skin AI with olfactory chemistry, I was able to solve my own frustration and build Diamandis — bridging high-end skincare diagnostics with personalized fragrance curation.
What it does
Diamandis diagnoses the user's epidermal metrics via YouCam Skin AI and curates a customized 2-step scent ritual:
- Epidermal Diagnostic Scan: Integrates the YouCam Skin Analysis API (v2.1) to measure 7 core metrics (moisture, oiliness, texture, pore, radiance, redness, and regional skin_type for whole face, T-zone, and U-zone). Real diagnostic zone masks are projected directly onto the user's portrait with interactive inspection.
- Step 1 · Skin-Prep Layering Base: A 100% biomarker-driven prescription (e.g., Ceramide & Squalane Lipid Balm for dry skin to lock in volatile top notes, or Oil-Balancing Niacinamide Gel Primer to keep crisp notes clean and prevent degradation).
- Step 2 · Curated Perfumes: Google Gemini matches the user's olfactory preference with their skin retention profile, recommending 3 catalog fragrances (EDT, EDP, Extrait), explaining note blooming, and estimating wear-time longevity in hours (e.g., ⏱ 7–9 hours).
- Personal Bottle Check: An optional feature allowing users to input any fragrance they already own to learn how its specific notes behave on their skin.
- Boutique & Judge Studio (
#/studio): A merchant management studio where judges can test with their own YouCam API key viax-youcam-keyheader and verify account units with zero consumption. - Luxury eCommerce Experience: A complete customer shopping journey with "Add to Bag", live badge feedback, a curated header bag drawer, and simulated boutique checkout.
Consumer Value: Ending the Era of "Blind Buying" Disappointment
Luxury fragrance is notoriously difficult to purchase online because paper blotters and scent descriptions cannot predict how volatile molecules will interact with a person’s unique epidermal chemistry.
- Eliminating Fragrance Regret: Diamandis removes the uncertainty from luxury fragrance purchases. By analyzing sebum levels, moisture barriers, and skin micro-texture through YouCam Skin AI, users know in advance how long a perfume will last on their skin (
⏱ wear-time in hours) and how the top, heart, and base notes will bloom. - Personalized 2-Step Performance Ritual: Rather than blaming their skin or the perfume, consumers receive a tailored prescription: an unscented priming base (Step 1) that balances pulse-point lipids, followed by complementary fragrances (Step 2) engineered to diffuse optimally across their epidermal barrier.
- Personal Bottle Validation: Consumers can input perfumes they already own into the "Personal Bottle Check" feature to discover why certain scents feel faint or harsh, learning how to prime their skin to maximize the longevity of their existing collection.
Retail & eCommerce Value: Transforming Beauty Tech into Revenue
For luxury beauty brands, boutique retailers, and department stores, Diamandis solves the single biggest bottleneck in online fragrance sales: high cart abandonment and friction in blind purchases.
- Higher Average Order Value (AOV) via Cross-Category Bundling: Diamandis creates an organic, scientifically proven bridge between skincare and perfumery. By prescribing a mandatory Step 1 Skin-Prep Layering Base alongside Step 2 Curated Perfumes, retailers naturally convert single-item perfume shoppers into multi-item skincare-and-fragrance bundle purchasers.
- Reduced Return & Dissatisfaction Rates: The primary driver of fragrance dissatisfaction is unexpected dry-downs or short longevity. By aligning fragrance concentrations (
EDT,EDP,Extrait) with the customer's skin retention profile before purchase, retailers dramatically decrease product returns and customer support inquiries. - Omnichannel In-Store & Online Utility (Boutique Studio): The built-in Boutique Studio (
#/studio) enables beauty advisors at retail counters to run instant skin analyses with customer consultations on tablets, input custom seasonal boutique catalogs, and instantly verify API credits using their store's YouCam key. - Higher Conversion Rates in Digital Retail: Interactive visual diagnostic masks from YouCam provide visceral "proof of expertise", turning skeptical shoppers into confident buyers who immediately click "Add to Bag".
How I built it
- Official Documentation: Developed strictly according to the Perfect Corp. Developer Introduction & API Reference.
- YouCam APIs (Perfect Corp):
- Skin Analysis API (
v2.1) (/s2s/v2.1/task/skin-analysis): Asynchronous task polling extracting 14 diagnostic biomarkers and SVG/PNG zone masks. - File API (
v2.0) (/s2s/v2.0/file): Multi-part temporary upload handling for user portraits. - Credit API (
v1.0) (/s2s/v1.0/client/credit): 0-unit balance verification for custom judge keys.
- Skin Analysis API (
- AI Curation: Google Gemini (
gemini-3.8-flash) with strict JSON schema response mode for structured recommendations. - Frontend: React 19, TypeScript, Vite, luxury CSS design system (Plus Jakarta Sans & Playfair Display typography), responsive split diagnostic view with skeleton image loading.
- Backend & Deployment: Node.js Express API deployed as a Vercel Serverless Function.
Challenges I ran into
- Connecting Facial Diagnostics to Pulse Points: YouCam Skin AI analyzes the face, but fragrance is applied to the neck and wrists. The challenge was establishing a valid biological bridge without making pseudoscientific claims. I resolved this by utilizing the facial U-zone (jawline and lower cheeks): it directly borders neck pulse points and serves as an accurate non-invasive proxy for systemic lipid production and stratum corneum hydration.
- Asynchronous Polling & Zero-Credit Resilience: YouCam's Skin Analysis API uses an asynchronous job model (
task_idwith status polling). I had to design a robust polling handler with clean error formatting (e.g., handlingerror_below_min_image_sizefor photos under 480px) and zero-credit fallback handling, allowing judges to test their own API keys viax-youcam-keywithout server restarts. - Strict Separation in AI Prompting: It was critical that Step 1 (Skin-Prep Base) remained 100% biomarker-driven and was never biased by the user's scent taste, while Step 2 (Perfumes) balanced both taste and skin retention. Enforcing this separation required building strict structured JSON schemas with Gemini.
- Asset Synchronization & Skeleton Loading: YouCam mask overlays are lightweight PNGs that download faster than high-resolution portraits, causing masks to temporarily float over empty space. I implemented an
onLoadsynchronization pattern with a shimmering skeleton loader so diagnostic masks reveal smoothly only after the base portrait has rendered.
Accomplishments that I'm proud of
- Pioneering a Novel Beauty-Tech Category: Creating the first functional bridge between clinical facial skin diagnostics and luxury perfumery, solving the real-world problem of fragrance evaporation and unexpected dry-downs.
- Deep Integration of Three YouCam APIs: Successfully orchestrating the Skin Analysis API (
v2.1), File API (v2.0), and Credit API (v1.0) in a unified production pipeline following official Perfect Corp. documentation. - Interactive Split Diagnostic Studio: Designing an intuitive interface where users can click any skin biomarker and instantly watch the corresponding YouCam diagnostic zone mask project onto their portrait with zero layout shifts.
- B2B Boutique & Judge Mode (
#/studio): Building a dedicated testing environment where judges and boutique staff can safely test custom YouCam API keys viax-youcam-keyand verify credit balances with 0 units consumed. - Complete Luxury eCommerce Experience: Delivering a fully responsive shopping journey — from portrait analysis and longevity estimates to interactive "Add to Bag" feedback and a curated header bag drawer.
What's next for Diamandis
- Real-Time Video Diagnostic via YouCam Web SDK: Transition from static portrait uploads to live real-time camera scanning, allowing shoppers to inspect their skin barrier and zone masks interactively in their browser or in-store mirrors.
- Environmental & Climate Adaptation: Integrating ambient weather APIs (temperature and relative humidity) alongside YouCam epidermal hydration to adapt fragrance wear-time predictions across changing seasons (e.g., dry winter air vs. humid summer heat).
- Personalized Discovery Sample Kits: Partnering with niche perfume houses to ship automated "Discovery Sets" — offering a 10ml travel-sized Skin-Prep Primer alongside 3 curated 5ml flacon vials so consumers can experience the ritual at home before purchasing a full-sized 50ml bottle.
- Smart Olfactory Wear-Log: Enabling users to log real-world wear time and track how seasonal skin changes alter their favorite fragrance performance over time.
Built With
- ai-curation
- beauty-tech
- ecommerce
- express.js
- fragrance
- gemini-api
- node.js
- perfect-corp
- perfumery
- react
- rest-api
- skin-ai
- typescript
- vercel
- vite
- youcam-api
