Gymnos
Built at Meta Global AI Developer Hackathon · Oct 26, 2026 · Remote
Logging workouts in a crowded commercial gym is broken. Tapping small buttons on a smartphone screen with sweaty or chalk-covered hands causes friction, disrupts rest intervals, and breaks workout focus. Generic voice assistants fail in noisy weight rooms because ambient music and clanking weights trigger false commands, while open-ended LLMs frequently hallucinate unsafe weight jumps. I built Gymnos to solve in-gym friction for real lifters. By combining Meta’s AI stack (Meta Voice ASR and Spark JSON extraction) with a human-in-the-loop confirmation safety net, Gymnos gives lifters a hands-free, noise-resilient workout session layer that logs sets in seconds and auto-regulates progressive overload without screen friction.