| App | How it coaches | Pushes workouts to | Price | Checked |
|---|---|---|---|---|
| 2PEAKSwiss adaptive training plans for triathlon, cycling and running that recalculate after every synced session, with two-way Garmin, Wahoo and Zwift sync.TriathlonCyclingRunningOther | Model-driven adaptive+ human coach | Garmin, Wahoo ELEMNT, Zwift | unverified | |
| AthleticaAdaptive endurance training plans with a conversational AI coach for running, cycling, triathlon and other sports.RunningCyclingTriathlonStrengthOther | Model-driven adaptive | Garmin, Wahoo ELEMNT, intervals.icu | unverified | |
| ICU CoachMobile app for intervals.icu users that generates daily workouts and weekly strategy from recovery data and training load.RunningCyclingSwimming | Model-driven adaptive+ human coach | intervals.icu | unverified | |
| TrainAsONERunning plans from a machine-learning model that recalculates the whole plan after each run, with a free tier and Garmin sync.Running | Model-driven adaptive | Garmin | unverified | |
| TrainerRoadCycling and triathlon training plans with machine-learning workout adaptation, fatigue flags and FTP estimation, played indoors or pushed to bike computers.CyclingTriathlon | Model-driven adaptive | Garmin, Garmin Edge, Zwift, Wahoo ELEMNT, Hammerhead Karoo, Apple Watch, Smart trainer | unverified | |
| TriDotAdaptive triathlon training platform with tiered subscriptions and an optional dedicated human coach.TriathlonStrength | Model-driven adaptive+ human coach | Garmin, Zwift | unverified | |
| XertPower-based cycling training platform that models fitness from every ride and recommends daily workouts or a forecast plan toward a goal.Cycling | Model-driven adaptive | Garmin, Garmin Edge, Wahoo ELEMNT, Hammerhead Karoo, Zwift, Rouvy, Smart trainer | unverified |
Model-driven adaptive training apps
Model-driven apps let a physiological or statistical model decide the next session: a fitness-fatigue model, a critical-power curve, or machine learning on your own history. TrainerRoad, Xert and Athletica are the typical examples. There is no chat; the method is documented, which tends to make these apps the most transparent in the index.
- apps match the rule
- 7
- not eligible for the shortlist
- 0
7 apps match this rule as of . 0 of them are not eligible for the shortlist (fewer than six dimensions scored from public evidence, or a score set by the site owner).
Shortlist: Model-driven
- ICU Coach
Mobile app for intervals.icu users that generates daily workouts and weekly strategy from recovery data and training load.
3.50 of 5 · 6/8 scored
- Xert
Power-based cycling training platform that models fitness from every ride and recommends daily workouts or a forecast plan toward a goal.
3.29 of 5 · 7/8 scored
- Athletica
Adaptive endurance training plans with a conversational AI coach for running, cycling, triathlon and other sports.
3.29 of 5 · 7/8 scored
Rule: Every app whose vendor documents a physiological or statistical model (fitness-fatigue, critical power, machine learning on the athlete's history) deciding load and session type, without a free-text conversation. Shortlist: the three eligible apps with the highest evidence-based mean score. Evidence mean is over the dimensions scored from public evidence; ties go to more dimensions scored, then the lower monthly price. Eligibility and tie-breaks.
AI Coaching Apps is made by the team behind YOUB. YOUB competes under the same rules and is shown in a shortlist only when a rule selects it. How this stays honest.
All 7 apps 7
Related categories
How this index works
- Facts, cited. Sports, AI level, thirteen coaching capabilities, devices, price: each from the vendor's own page, linked and dated.
- Scores, from evidence. Eight coaching dimensions, 1 to 5, each citing a public page. No evidence, no score: a blank is not a low mark.
- Rules, not opinions. Every shortlist is the printed output of a rule over the data, recomputed on every build. Nobody picks a winner.
Every fact on this page links to the vendor page that states it on the app's own page. Scores follow the published rubric. The compare page shows all eight scored dimensions side by side.