GymLog - Open-source voice + AI workout log for lifting and CrossFit

Hi Product Hunt 👋 I built GymLog so my coach and I could log and review our daily training together, far more easily than passing a spreadsheet back and forth. Beyond the two of us, I want it to help any coach manage training data for a large roster of athletes and leave the traditional spreadsheet behind. Over 90% of GymLog's code was written with GPT-6 Astra, which is why I'm launching it as part of the GPT-6 Astra Challenge. GymLog is a SwiftUI + SwiftData iOS app for people who do both conventional lifting and CrossFit-style WODs: • Strength: per-set load and reps, supersets, drop sets, rest timer. • WODs: AMRAP, For Time, EMOM and intervals, with background phase alerts and a timer that resumes after the app is closed. • Honest PRs: scores are compared only when movements, loads, time caps and scoring rules match, so scaled or modified workouts don't distort records. • Coach ↔ athlete: exchange plans and results as share files, no re-typing. • Your data: Excel import, CSV export, full backup/restore, on-device InBody report recognition. 240 bilingual (EN/中文) exercises built in. • Voice + AI (bring your own API): describe a session out loud to build or change the plan, and get AI reviews afterwards — routed through a Cloudflare Worker relay you deploy, so provider keys never ship in the app. It's MIT-licensed and currently a source distribution: build with Xcode + XcodeGen (iOS 17+). Energy numbers are MET-based estimates, and the AI reviews don't assess technique. I'd especially love feedback from coaches who track many athletes: what would it take for GymLog to replace your spreadsheet?