2025 – 2026 · AI-Assisted Development
Multi-tenant gym management platform for the Australian fitness industry
A production-grade, multi-tenant SaaS platform — designed, architected, and built almost entirely through AI-assisted development as a solo developer
Gym SaaS is the most technically ambitious project I have built — a full-stack SaaS platform spanning four deployed sub-applications and a shared Turborepo monorepo. It was developed primarily using Claude (Anthropic) and Cursor as AI coding assistants, which dramatically accelerated the scope and quality of what I was able to ship alone.
The platform covers the complete lifecycle of running a gym business: member management, class scheduling and QR check-in, multi-product billing via Stripe, document compliance and digital signing, multi-location support, personal training management, and a React Native mobile app with biometric authentication. Each of the four sub-applications shares a tRPC v11 API layer, a Drizzle ORM schema over Supabase PostgreSQL (35+ tables with row-level security), and a cross-platform @gym/shared package.
AI-driven development was not just a productivity tool — it was the core methodology. AI tools designed the database schema, scaffolded all 27 tRPC routers with end-to-end type safety, implemented complex multi-tenancy and RBAC patterns, wrote React Email templates, and maintained consistency across a large codebase. Features that would typically require a small team were shipped in days.
Staff-facing dashboard — member management, class scheduling, QR check-in, multi-product billing, document compliance, audit logs, and full RBAC.
Next.js 15 · tRPC · Drizzle ORM
Member-facing web app — class booking, product shop, Stripe checkout, digital document signing, notifications, and profile management.
Next.js 15 · tRPC · Stripe
Public-facing marketing site for the platform with feature showcases and onboarding flows for gym operator acquisition.
Next.js 15 · Tailwind CSS v4
Native iOS/Android app — class booking, QR check-in, product shop, biometric login (Face ID / Touch ID), dark mode, and custom tab navigation.
Expo SDK 54 · React Native · NativeWind
This project was built almost entirely through AI-assisted development. Claude and Cursor were used as primary development partners — not just for autocomplete, but for architectural decisions, full feature implementations, database schema design, API layer scaffolding, and code review. Cowork handled file management, cross-referencing documentation, and multi-step workflow automation.
Architecture decisions, database schema design, all 27 tRPC routers, email templates, RLS policy review, CLAUDE.md conventions, and multi-step workflow automation across the monorepo
Real-time pair programming, inline edits, multi-file refactors, and rapid feature scaffolding — used for the majority of day-to-day implementation
Treating AI as a senior engineer: precise context windows, explicit constraints in CLAUDE.md/AGENTS.md, task-specific agents, and iterative verification loops