Peloton Patents Point to a Future Where AI Workouts Adapt in Real Time

New patent applications hint that Peloton may be looking to use AI to change workouts and instructor coaching based on a member’s performance while a class is underway

Peloton has spent the past year making artificial intelligence a bigger part of its fitness platform. Newly published patent applications by the company may offer insight into where those AI investments are headed.

Two applications from Peloton Interactive published last week describe technology that could modify workouts in real time based on a member’s performance, including changing the structure of a class and generating new instructor voice cues while the workout is underway.

The applications were published by the U.S. Patent and Trademark Office on August 27, 2026, though both were originally filed on February 27, 2025. Peloton chief technology officer Francis Shanahan is listed as the inventor. The filings are patent applications rather than granted patents, and Peloton hasn’t announced commercialization plans yet.

One application, titled “Real-Time Modification of Workouts Within a Connected Fitness Platform,” describes a system in which a workout can take different paths based on data collected while a member is exercising. The system can insert, remove, shorten, lengthen or replace workout segments depending on a user’s performance.

Peloton describes existing connected fitness workouts as largely “static and inflexible” once they begin, even when the class itself was recommended specifically for the user. The proposed technology would instead create what the filing characterizes as a “choose your own adventure” workout, with different branches available depending on factors including heart rate and other performance metrics captured by exercise equipment or a connected wearable.

A companion application that Peloton filed takes the concept a step further.

“Real-Time Modification of Audio Content for a Virtual Coach Application” describes using a large language model to generate altered coaching instructions based on those same signals. The system could compare a member’s performance with a target and produce a new instruction in response. It also contemplates using a speech synthesis model associated with a particular instructor and trained using classes taught by that instructor.

In one example contained in the filings, a runner completes an interval several seconds faster than intended. The system responds during the recovery period with an instructor-style cue telling the runner to slow the next interval. In another, the system determines that a strength set was too difficult, reduces the final set to six repetitions and generates a corresponding coaching instruction.

AI-Powered Adaptive Fitness

The company launched Peloton IQ in October as an AI and computer vision system designed to provide personalized guidance across the Peloton platform. Its current capabilities include personalized workout plans, performance estimates, weekly insights and recommendations, along with more advanced strength features such as rep tracking, form feedback, suggested weights and a workout generator on certain equipment.

Nearly half of Peloton’s active members had engaged with Peloton IQ insights and recommendations by the end of the company’s fiscal second quarter, according to Peloton.

Much of that personalization, however, still happens around choosing existing workouts: Peloton can determine what a member should take, establish an appropriate performance target or analyze what happened afterward.

The patents describe a potentially different model, where personalization increasingly happens during the workout itself.

An instructor could effectively provide the framework for a class, while software determines which version of that workout an individual member receives. AI-generated audio could theoretically preserve the instructor’s coaching style as the underlying workout changes.

In his 2026 annual letter, CEO Peter Stern wrote that Peloton sees AI as a way to move from “counting and tracking what you did to personalizing and coaching you on what to do.” The company has argued that its workout data, combined with information members choose to share through outside integrations, can give it an advantage in building personalized fitness experiences.