Hi PH 👋 I built Shindo because I got tired of the integration problem: my watch had a readiness score, one app had my fitness and fatigue curves, and my training plan lived in a spreadsheet that stopped resembling my life by week three. Every morning I was doing the synthesis myself, badly, at 6am. Shindo does that synthesis. It reads Apple Health — whatever device wrote the data — computes readiness with the four drivers behind it, models training load per sport, and builds the week. When you miss sessions or swap a run for a ride, it counts what you actually absorbed and re-flows the block instead of leaving you with a plan for an athlete you no longer are. The part I'd most like feedback on is the constraint I built it around: there's a language model in the app, and it is not allowed to do arithmetic. Every number is computed deterministically on-device first; the model receives finished numbers and writes sentences about them, and if its text disagrees with a computed value the computed value overwrites it. There's exactly one exception — the weekly plan, where the model may propose a week that a 19-rule validator then accepts or rejects against limits derived from your own data. If it rejects twice, you get the engine's own week and never see a failure. The whole methodology is published, constants included: https://shindo-app.com/methodology It's free for readiness, the plan, the load charts and workout detail. Pro adds the AI interpretation and the forward-looking models. Honest limits, so nobody wastes a download: iPhone only, no Android, it can't push workouts to your watch, and it doesn't record sessions — it reads what your device already recorded. If you don't own something that tracks sleep and HRV, it has nothing to work with. Happy to answer anything.

Shindo Endurance Training

Your first triathlon, on the watch you already own

Your first triathlon, on the watch you already own