Dropshot Coaching

Context

A platform where tennis players film their strokes and receive AI-generated technical analysis and personalized coaching. Mobile app for players, web backoffice for admins.

Scope

Full development of the mobile app, the admin web app and the whole content chain: videos, generated copy and translations, frame extraction for thumbnails, lesson recommendations. The video analysis (player angle, recommendations based on tennis metrics) was built by the IA LAB studio, in close collaboration throughout the project.

Product decisions

• Designed the 4-step coaching pipeline (AI analysis, frame extraction, coaching generation, delivery) with per-step retries and a cron safety net
• Coaching text generated by Claude with structured output, using the player's last 3 sessions as context
• Build-vs-buy R&D on multilingual video dubbing, documented with the trade-offs so the client could decide

Highlights

• AI video analysis pipeline in production, with push notifications on completion
• Lesson recommendations powered by a vector database
• Mobile subscriptions via RevenueCat
• 3 languages (FR, EN, ES), with the coaching generated in the user's language
• Monorepo Turborepo: Expo app, React backoffice, Convex backend, CI per app
• Mobile app currently rolling out to the stores

Stack