Reservine
Founder-built booking SaaS with a deep, multi-provider AI backend.
- Angular 21
- own UI library
- Nx
- Laravel 12
- multi-tenant
- Stripe Connect
A multi-tenant reservation platform for any business. I founded it and own the Angular 21 Signals frontend and its 73-component UI library; the product runs 16+ AI services on a 19-provider LLM abstraction with structured output. Built with co-founder Martin Foltýn, who owns the Laravel backend.
The product
Reservine is a multi-tenant booking SaaS — one platform a salon, clinic, gym, or studio runs its entire reservation flow on. Every tenant is data-isolated (stancl/tenancy), with its own calendar, customers, and Stripe Connect payouts.
I co-founded it, and I'm precise about who built what. It's a two-founder build: I own the Angular frontend, the in-house UI library, and the product and UX direction; my co-founder Martin Foltýn owns the Laravel backend, including the AI services described below. Neither half is solo work dressed up as the whole.
It's live at reservine.io.
The frontend
The frontend is Angular 21, signals-first throughout — server state lives on TanStack Query, everything else on signals.
I didn't reach for a vendor component framework. I built our own: a 73-component UI library on an Nx monorepo over Tailwind and DaisyUI, so the whole product shares one consistent, owned design system instead of fighting someone else's.
The piece I'm proudest of is the calendar. A booking product lives or dies on its timeslot grid, so I wrote a GSAP-driven virtual-scroll calendar engine that renders only what's on screen and offloads the heavy layout math to Web Workers. The main thread stays free, and the grid stays smooth even when a tenant scrolls across thousands of slots and resources.
The AI backend
The AI layer is unusually deep for a booking tool — and it's Martin's. He built it on a 19-provider LLM abstraction: a single provider enum, so the product isn't married to one vendor.
On top of that sit 16+ AI services with structured output (Instructor-PHP), driving roughly a dozen customer-facing features: no-show and lifetime-value prediction, dynamic pricing, upsell and smart-slot recommendation, review and cancellation insights. Every call is logged per tenant.
What I built here is the surface: the admin dashboard that turns that per-tenant logging into AI usage and cost observability — the view an operator actually reads. The services and the provider abstraction behind it are Martin's; the control surface over them is mine.