Computer VisionVite / ReactReal-time UX
Velo Coach
AI pose coaching in the browser
VELO brings AI gym coaching to the browser with pose detection, rep counting, and session feedback for athletes and experimenters.
- React
- Vite
- WebRTC / CV
- Vercel
Client
Quezt Labs
Industry
Fitness / AI
Key results
Pose + Reps
AI
Browser
Client
Vercel
Deploy
The challenge
Gym coaching apps usually require native installs and heavy ML pipelines. We wanted to prove browser-based pose coaching with acceptable latency for demos and pilots.
Our approach
- Camera access + frame processing in the client
- Rep counting and form hints surfaced in a minimal UI
- Vite for fast deploy cycles on
velo-coach.vercel.app - SEO/meta for discoverability as an R&D showcase
The solution
VELO Coach runs as a lightweight web app — open the site, start a session, get live feedback without an app store round-trip.
Technical notes
- Prefer progressive enhancement when camera APIs are blocked
- Keep models and inference paths measurable (FPS, drop frames)
- Separate demo mode from production coaching flows
Impact
- Public showcase of Quezt Labs AI + fitness crossover
- Reusable patterns for vision features in client MVPs
- Validated that Vite + Vercel is enough for interactive CV experiments