AptixLabs ships on a deliberately small, modern stack so every product is fast to build and cheap to operate. The studio standardises across projects so partners onboard in days rather than weeks, and the infrastructure cost stays under a few hundred dollars a month even at meaningful scale.
The frontend layer
Marketing surfaces and the coach portal run on React 18 with TypeScript and Vite. Mobile is React Native with TypeScript, shipping to both Apple App Store and Google Play from a single codebase. Motion is Framer Motion — the studio uses it sparingly, only when an animation does communicative work.
The backend layer
Google Cloud is the default. Cloud Run hosts stateless services, Firestore is the primary database, Firebase Auth handles identity, and Vertex AI is the home for any model inference that isn't handled by a partner API. Voice and conversational AI come from ElevenLabs. CI is GitHub Actions. Static hosting is Firebase Hosting with prerendered HTML for SEO.
What we deliberately don't use
- Kubernetes — Cloud Run handles every workload at the studio's current scale
- A separate Redis or pub-sub layer — Firestore is the message bus
- A bespoke design system — we lean on tokens + Tailwind-shaped utility classes
- A separate CDN — Firebase Hosting + Cloudflare in front is enough
- GraphQL — REST + Firestore listeners cover every read pattern we have
When something on this list stops paying for itself we replace it — that's how Postgres got demoted to "only if you really need joins". The stack is a living document.
