Time caps from data, not gut feeling.
A programming engine that writes your week the way you would — balanced across the cycle, in three tracks, ready for the whiteboard. Every time cap is derived from how long workouts like it actually take, not from how long they feel like they should.
Founding members: 50% off for life. No card, no spam.
What comes out
One workout, three tracks, generated together from the same stimulus — so your newest member and your fire-breather train the same piece.
Why it isn't a chatbot with a workout prompt
Calibrated time caps
Per-movement paces learned by non-negative least squares over labelled workouts, with a fatigue term — 155 wall balls is not 155 × your fresh pace. Mean error ±3–5 min on the benchmark set.
Statistical, not generative
Movements, rep schemes and loads are sampled from the empirical distributions of a 1,742-workout corpus. The language model never touches a rep, a load or a cap — it only reads your request and formats the output.
A week that holds together
Cycles are built against statistical archetypes with adjacent-day constraints, so movement patterns and energy systems stay balanced. Variety by design, not by accident.
Three tracks, one stimulus
RX, Scaled and Beginner are generated in the same pass from the same intent — not scaled down afterwards by whoever is on the floor.
Where a cap comes from
Take the workout above. Every comparable couplet in the corpus is scored for movement pace, volume and fatigue; the cap lands where the field actually finishes — not on a round number someone liked.
Guessing a cap at 12:00 would have timed out roughly a third of the room. Guessing 20:00 would have left everyone standing around.
Stop guessing your time caps.
Built by a coach who runs an affiliate. Tested daily on a real gym floor.