Commerce · MACH
FeminineXP — AI-first D2C beauty with AR try-on & fit agent
A headless D2C beauty platform that pairs on-device AR with a conversational AI Fit Agent, built multi-tenant so the same codebase powers multiple brands with a single config switch.
The challenge
A sanctuary experience for beauty commerce
Built for a beauty brand that wanted more than a storefront — they wanted a sanctuary. Customers explore looks with a Virtual Vanity AR mirror (on-device, no data leaves the phone), ask the AI Fit Agent about ingredients and skin types, and check out with confidence. The platform is multi-tenant so brand #2 is a config flip, not a rebuild.
What we delivered
- On-device AR try-on with zero privacy risk
- Conversational AI that lifts buyer confidence
- Multi-tenant — brand #2 ships without a rebuild
- PWA + native on iOS and Android from one codebase
- Headless storefront with no-code CMS admin
- Accessibility-first, reduced-motion aware
Features
What we built
Six core capabilities that make the product what it is.
Virtual Vanity AR Mirror
On-device TensorFlow.js model for real-time product try-on. Shade matching, skin-tone detection, and beauty filters — zero cloud data sent.
AI Fit Agent
LangGraph-powered conversational agent grounded in product data. Answers "is this right for my skin type?" with cited, trustworthy reasons.
Conversational Discovery Guide
A guided quiz that narrows from full catalog to a curated recommendation set in 3–5 turns — no browsing paralysis.
Headless Storefront
Next.js frontend with persisted cart, optimistic UI, variant selection, and a Puck CMS admin for non-technical merchandising.
Multi-tenant Architecture
Per-tenant theming, catalog, pricing, and content — activated by config, not code. New brands go live without a rebuild.
Mobile PWA + Capacitor
Installable on iOS and Android via Capacitor, with offline browsing and camera AR access — no app store dependency for the PWA path.
Stack
Built with the right tools
No dogma about tools. We assessed what the product needed, then chose the stack that delivers fastest and stays maintainable long-term.
Ready when you are
Build something like this?
Tell us what you're working on. We'll map the right approach and stack for your context.