Hello, I’m new at Tainerroad software. I started to use it right a week before introducing the new AI model for everyone. I had a 10-day stage race (Zwift) that included several mountain stages, which I completed mostly out of the saddle. Due to my performance out of the saddle being much higher than in the saddle, the first thing that TR did was a significant overestimation of my FTP. That was not an issue, I thought, setting the FTP manually. Then the new AI occurred, and FTP overestimation was exacerbated even more. Much more. But now I’m supposed to use AI-detected FTP in order to be able to use new features. It’s not a big deal, though. The main issue for me is that AI doesn’t really use that AI-detected FTP for determining which workout to prescribe. It looks into what I’m capable of for certain time periods, like 1 min, 5 min, etc. Correct me if I’m wrong. For now I get the workouts that “feel just right” only if I perform them out of the saddle. Otherwise, there is no way I can do them in the saddle. For example, there was a workout yesterday called “Dashuik”: 4x5’@109% with 10m recovery in between. So, I managed to do the first interval. The second one and others I had to do out of the saddle after the first 2 minutes of them. Basically, I’ve done this workout, and AI has no clue that I’ve failed it because the power and duration targets were reached. That situation makes me ask questions.
How can I address this problem?
Is this even a problem?
Is this tug of war between the high performance during races and not being able to keep it up during workouts, marking them as “max effort,” going to lead AI to prescribe me proper workouts eventually?
Do you consider adding cadence as a useful metric for AI in the future? Like, for detection if someone is struggling during a workout or if someone is working out of the saddle, these power numbers shouldn’t be used for performance assessment.
Sorry for my English. It is not my native language.