WHOOP’s AI Rebuild: My Memory and Check-Ins

WHOOP has sold itself for years as a recovery tracking band rather than a sports watch. There is no GPS and no map screen. Its whole value sits in how it interprets the data.
In May 2026 the company announced its largest rebuild of the AI layer since WHOOP Coach arrived. The two features are My Memory and Proactive Check-Ins, rolled into the app gradually over the summer. By the end of the holidays they should be live for every user.
My Memory: context that carries over
The problem with the AI Coach was simple: every conversation started from nothing. Tell the app about a 70.3 in three weeks, then mention a calf that needs rehab a week later, and the system would not join those two facts up when it produced the next training suggestion.
My Memory is a place to review, correct and delete the context the AI has built up about an athlete. A race goal, an injury, a holiday, a change of training plan all live there. Instead of retyping any of it at the start of each conversation, one entry gets edited and the recommendations are meant to carry it forward.
Proactive Check-Ins: the app asks first
The second feature reverses who starts the conversation. Until now the athlete opened the app and asked whether today was a good day for threshold work. Proactive Check-Ins are meant to offer the recommendation at the right moment on their own: prioritising sleep before an important race, or adjusting the plan after a flight across time zones.
The package also includes a rebuilt Journal that works by voice. Instead of tapping through a list of behaviours, the user records a note. WHOOP has also announced medical record integration (EHR Sync) and on demand doctor consultations, though those are health features aimed squarely at the US market.
Heart rate accuracy
WHOOP has promised further work on the heart rate algorithm, with the emphasis on the activities where it performed worst: strength training, functional work and HIIT. For a triathlete what matters most is measurement in the water and on the bike, and there the band has historically trailed a chest strap.
There is no independent accuracy test of the 2026 update yet. In an August comparison of screenless bands, DC Rainmaker noted that wearing the band on the upper arm gives accuracy close to a chest strap. That test carries no separate verdict for swimming or riding.
Not a replacement for a sports watch
WHOOP will not stand in for a watch on race day. It shows no running pace and no live bike power. It wins elsewhere, in reading the load between sessions: Strain, sleep quality, and now context the AI actually remembers from one week to the next.
For an athlete training with a Garmin or a Coros who wears the WHOOP mainly for recovery, the update earns its place. Less time goes on telling the app something it has already been told.
Data is one thing, a plan is another
This is where the boundary sits. WHOOP reports how the night went and how much load the body will take today. It does not build a triathlon plan. It does not know there is a half distance race in six weeks and that swimming is the weakest of the three.
That second part is what Triathlify does, an app built in house, so the recommendation comes with that attached. It pulls data from Garmin, Strava and Apple Health and recalculates the plan every week. After two short nights it cuts volume and intensity by itself rather than asking for missed sessions to be made up. Integrations with WHOOP, Polar, Coros, Suunto and Oura are still in development. Periodisation covers everything from sprint to full IRONMAN, and the first seven days are free.
In Europe a WHOOP subscription costs 199 EUR a year for the One plan and 264 EUR for Peak. The top Life plan is 399 EUR and includes the WHOOP MG band with ECG measurement. The band is included in the price rather than bought separately, and the hardware is sold only through the manufacturer’s own site.
The new AI features arrive at no extra cost on every plan.
