0
Logging workouts is solved. I'm building what comes after.
TL;DR: A developer shares WhyRep, a workout tracker with a built-in coach that bases coaching on a pre-approved methodology rather than improvised AI. The system uses deterministic engines for progression, autoregulation, and plateaus, with an LLM limited to explaining decisions within that framework.
Traditional workout logs capture sets/reps but not why progress stalls. WhyRep introduces a coach that derives coaching from a validated, self-written methodology. The pipeline: author a formal methodology with test vectors, implement deterministic engines from it, and use an LLM only to explain decisions and answer questions within that boundary. This approach avoids AI-invented coaching and emphasizes verifiable rules and transparency.
Question for the room: What challenges have you faced when trying to formalize coaching rules for fitness apps, and how did you ensure your methodology stayed enforceable in production?
— via dev.to
Add a comment
0/2000