This AI Trainer App Wants to Make You a Faster Cyclist

For those with no competitive goal in mind, who are just interested in building their fitness, there’s also the TrainNow option in which TrainerRoad allows you to choose from a selection of daily workouts from three categories: Climbing, Attacking, and Endurance.Adaptive Training may be smart, but it’s still not smart enough to eliminate the need for ramp tests to establish your baseline “functional threshold power” (FTP). Determined on a scale of 1 to 10, they are calculated using three methods: machine learning, the company’s already extensive set of anonymized data gleaned from millions of completed workouts by other athletes, and your own recent workout performance.TrainerRoad's software can sync to just about any smart trainer or the power sensors on your bike.TrainerRoad’s Adaptive Training appealed to me. Impatient with Erg mode, I redid the ramp test in manual mode, which I had to cut short, because I spent my allotted training time trying to fix the bug.The Adaptive Training workout programs change over time as the algorithm measures your improvement and learns your limits.The first few intervals after the ramp test felt a little too easy, which was no doubt a direct result of my incomplete ramp test. Because I don’t have a power meter on the fat bike I ride in winter, I used TrainerRoad’s option to input a “rate of perceived exertion” (RPE), a self-reported measure of the intensity of your effort, which you plug in using a scale of 1 to 10.Bundled up in multiple layers, big mitts, and cycling boots, I made a full-on effort to follow the scheduled interval training, a 70-minute workout titled “Seneca Rocks” in my “Sweet Spot” training zone, which contains intervals in the 88 to 94 percent range of my FTP.