Methodology — in the open.
Most AI triathlon coaches decide your week without showing the math. TriPaced does the opposite: every plan change carries the rule that fired, every number you see on the dashboard is the number the engine actually uses, and the methodology behind it is right here. Three pillars, no black box.
The Banister model
CTL, ATL, TSB — fitness, fatigue, balance. The 50-year-old impulse-response model that drives every plan change, exposed live with the numbers actually feeding the engine.
Physio guard-rails
Twelve hardcoded physiological limits the engine refuses to cross — drawn from the sports-science literature, not from product opinions. Why your CTL ramp gets capped before injury risk does.
Multi-discipline load
80 TSS on the run breaks your legs. 80 TSS on the bike does not. Most triathlon AI coaches sum them — TriPaced keeps swim, bike, run and strength TSS apart, and balances them per discipline.
Why we expose all of this
The underlying science is not proprietary. Eric Banister described the impulse-response model in 1975. Andrew Coggan formalised TSS for cycling power in the early 2000s. Joe Friel's Triathlete's Training Bible has codified the long-course guard-rails for two decades. Pretending TriPaced invented any of this would be dishonest.
What we did invent is the composition: how these models talk to each other in a single adaptive engine, where each rule's authority is documented, and where the athlete can see — and argue with — every call.
Opacity is fragile. When the AI coach is wrong and you cannot tell, you stop training. When the AI coach is wrong and you can see why it is wrong, you fix the input and keep training.
What the engine reads, in order
Every plan change runs through the same stack, in this order:
1. Banister state. What does your current CTL, ATL and TSB say about your readiness for the week as planned? Trajectory toward race-day TSB target?
2. Per-discipline balance. Is the load distributed correctly across swim, bike, run and strength for this phase? Is the run TSS share inside its safe band?
3. Physio guard-rails. Would this week violate a hardcoded limit (CTL ramp, TSB floor, long-run cap, taper minimum, swim frequency, brick density)? If yes, the engine yields to the rule — not the other way around.
The result is a plan you can read and a plan you can challenge. That is the whole product.
The three pillars in one paragraph each
Banister model. A 42-day exponentially-weighted average of your TSS (CTL = fitness) minus a 7-day average (ATL = fatigue) gives you TSB. Race-day target lives at TSB +5 to +15 for an Ironman, +10 to +20 for a 70.3. Every dashboard shows the three numbers; every plan change cites which of them moved.
Physio guard-rails. Long run capped at 2h30 for Ironman prep (the three-hour rule, Friel). Taper at minimum 3 weeks for Ironman, 10 days for 70.3. CTL ramp at 5 to 8 TSS per week. Swim at least 3 sessions per week in Base/Build/Peak. The engine will never generate a week that breaks any of these — even when you ask.
Multi-discipline load. Bike TSS, run TSS, swim TSS and strength TSS are tracked separately. The same TSS number absorbs into the body differently per sport. TriPaced reads per-discipline CTL and balances the week so no single sport ramps faster than its safe rate.
Frequently asked
What does "glass-box" mean for a triathlon training plan?
Every plan decision carries the rule that fired next to it. When TriPaced moves your Tuesday session to easy, it tells you why — TSB below the floor, or CTL ramp would exceed the safe limit. You can argue with the model when it disagrees with how your body feels. A black-box AI coach gives you nothing to argue with.
Why does TriPaced publish its methodology instead of treating it as IP?
Two reasons. First, the underlying science (Banister 1975, Coggan's TSS, Friel's training-bible guard-rails) is public — pretending it is proprietary would be dishonest. Second, transparency is retention: athletes who understand why a session changed trust the plan and keep training. Opacity is fragile; trust is not.
How is this different from TriDot, Humango or Athletica?
All of them are black-box AI coaches: they decide your week without showing the math. TriPaced exposes CTL, ATL and TSB live, names the physiological guard-rail that fired on each plan change, and publishes its methodology. We are also long-course only (70.3 and full Ironman), where the others are multi-sport or multi-distance. See the full comparison.
Do I need to understand CTL and TSB to use TriPaced?
No. The plan works on autopilot if you want it to. The methodology is there so you can look inside when something feels off. Most athletes only read the methodology pages once they want to argue with the engine about a specific call — that is exactly the point.
Last updated: 2026-06-22
Apply for beta access.
Free during beta. We review every application by hand and reply within 7 days. Long-course only — 70.3 or full Ironman.
Apply for beta →