A solo developer has launched RivalRep, a browser-based application that transforms push-up and pull-up training into a competitive duel against your own history. The project addresses a common problem in fitness tracking: the mental barrier where users quit prematurely because a static counter lacks the psychological pressure of a rival. By leveraging the device camera for real-time rep counting, the app introduces dynamic opponents that push users past their usual stopping points.

The Ghost Mechanic and Adaptive AI

The core feature is the 'Ghost' mode, which replays a user’s personal best set in real time. Unlike simple numerical targets, this feature provides visual and temporal context, showing exactly how a past self performed each rep and pause. This creates an immediate sense of pacing and urgency. Additionally, the app includes adaptive bots that analyze past sets to identify weak points. These bots intentionally accelerate during the specific intervals where the user historically fades, providing a tailored challenge that normal counters cannot offer.

Privacy-First Architecture

From an infrastructure perspective, RivalRep is notable for its privacy-centric design. All pose detection runs locally on the user’s device, meaning no video or image data is ever uploaded to a server. This eliminates bandwidth costs for the developer and removes privacy concerns for the user. The application is fully browser-based, requiring no app store download. It works on iPhones, Android devices, and laptops via webcam, though users can add it to their home screens for an app-like experience.

Multiplayer and Training Modes

The platform supports live duels via shareable links, allowing friends to compete head-to-head without needing accounts. The system offers three distinct challenge modes: Classic for maximum reps in one minute, Survival for maintaining a pace of one rep every five seconds, and Time Trial. An adaptive training plan automatically adjusts based on user feedback and progress, ensuring the difficulty scales appropriately. The developer emphasizes that accounts are only necessary for saving backups or engaging in multiplayer features, keeping the barrier to entry low.

Key Takeaways

  • RivalRep uses local pose detection to count push-ups and pull-ups without uploading video data.
  • The 'Ghost' feature replays personal bests in real time to create psychological pressure.
  • Adaptive bots accelerate during intervals where the user historically struggles.
  • The app is browser-based, free, and requires no account for single-player training.
  • Live duels allow friends to compete via shareable links without downloading an app.

The Bottom Line

This is a clever use of local inference to solve a psychological problem rather than a technical one. By keeping the stack simple and the data local, the developer has created a highly scalable, privacy-friendly tool that feels personal without needing a massive backend infrastructure.