Bezh

Beauty salons in Iran had no unified way to list services, take bookings, and let customers see a look before committing to it - each salon managed appointments manually, and customers had no way to preview a hairstyle or color on themselves before walking in.

A multi-tenant marketplace where each salon manages its own staff, services, hours, and bookings, layered with an AI Mirror: customers upload a photo and get an AI-generated preview of a hairstyle, hair color, or makeup look, gated behind a per-salon subscription that salons purchase through Zarinpal.

Full-stack build: the multi-tenant schema (organizations, staff permissions, services, appointments), the AI Mirror pipeline (private photo storage, AI-backed generation, shareable results), the subscription and payment flow, and a Tiptap-based articles CMS for the platform.

Next.js 16, Prisma over PostgreSQL, NextAuth v5, next-intl, Tailwind CSS v4, Tiptap for the articles editor, Zarinpal for payments, and an AI-backed image generation pipeline for the AI Mirror, with a mock provider swapped in for local development.

A warm, cream-and-brown palette distinct from the platform's own admin surfaces, with the AI Mirror's uploaded photos deliberately kept out of public storage entirely - readable only through an authenticated, session-scoped API route rather than a public URL.

Built the subscription-gating logic so the AI Mirror only becomes visible on a salon's page once that salon has purchased access, rather than exposing it platform-wide by default - plus a moderation layer in platform-admin for reviewing organizations, payments, and AI Mirror subscriptions across every salon.

A working multi-tenant platform live at bezh.beauty, with salon onboarding, booking, and the AI Mirror preview pipeline functional end-to-end - currently onboarding its first salons.