This is super interesting, I’ve never look at training this way.
My last two hard workouts where Blackcap and Emma which I was able to complete but gave it a Very Hard rating.
Blackcap was so borderline that I really had to think about that workout before rating because it was the Over section that I found Very Hard.
Liking at what you posted and the workouts Three Sisters and Dinara, I can see how those workouts are similar I thinking to give me another chance in that zone before moving forward.
@Nate_Pearson So why have AI FTP detection at all and just let the plan progress, since the system isn’t reliant on FTP? I know this seems scary to many people because we rely on FTP for pacing during races and to benchmark “bike fitness” but it seems it’s the center of the confusion. Maybe we could have the system predict FTP twice a year and it more accurately reflect our true FTP rather than a modified FTP every 6 weeks. Just a thought.
TR should just give people the most optimistic FTP’s possible and then just give them the actual training they need. Monkey brains need to be fooled. The monkey brain ego needs to be stroked with some good FTP points. If the rider insists on an FTP higher than is realistic, the Ai should just humor them.
I think it’s nice as a process check. But it’s more of a training score. I imagine once they launch the 20 and 60 minute power estimates, they could easily rebrand it as such.
I don’t agree. Reality is that, to date, the cycling community at large regard FTP as a training score and really never understood what it is or isn’t. Nor do they really care…unless their number ‘gets worse’.
Best I would advise, is that if you do understand these things, just don’t worry about it. Pick a real measure of performance, hang your self-worth on that.
Wow - I just checked my calendar and it is there as well
I remember a while ago asking my (then) coach whether such a thing as a NP Power Duration Curve existed. He said he didn’t think so. IT will be interesting to see how this new feature is used.
It will be really useful for our athletes when they want to look at how upcoming workouts compare to the work they’ve been doing recently – especially after FTP changes.
I was happy to read this and took it as an explanation for the over-estimation of my AI FTP detection. However after having a workout my conclusion is that the FTP value is taken into recommendation.
Please follow my situation:
Had a manual FTP setting of 275W, sometimes barely able to complete workouts. While my FTP was 275W TrainerRoad estimated my FTP at 290W. Knowing I’m barely able to complete workouts I ignored this.
The workout proposed today was a VO2max training with 15 sec intervals at 344W.
Reading the statement that the FTP value is not being used for workout recommendations I was triggered and accepted the AI FTP estimation.
Started the workout today and noticed in Zwift that the 15 sec intervals have increased to 363W.
After 3 blocks of max effort and dizziness I had no other option than to reduce the effort level with 5%. This has put the 15 sec intervals back at 344W which was still hard, but possible to finish.
I reduced my FTP manually to 270W which I think is about the right value.
With this behavior I don’t understand that the FTP parameter is not taken into account for setting the workout intensity. Out of my experience this just doesn’t seem to be the case. Did I misunderstand your statement? Can you elaborate further or point me to some documentation?
In my case the FTP prediction before the new AI model seemed more accurate.
Another option would have been to mark the effort all out. I’m assuming it gave you a “why did you fail” survey. If you answered those, the system would have adapted your workouts down. Just another option
Curious. Why would the system suggest a level 4 training for a lower FTP and a level 1 training for higher FTP? I’m getting back into training so I am just letting the Ai system guide me back to better fitness. I’ve been trying to figure out this Ai stuff and why I can’t remove the black boxes on the calendar and why the system keeps altering my chosen workouts. All this stuff just hurts my brain matter and I need it for something other than this, so I am letting Ai recommend the workouts for me (including FTP) and just doing what pops up on the calendar.
I entered “Max effort” but I didn’t get a survey. I only get one when I select “I did not pass”, but as I have finished the workout - even though with lowering the intensity - it doesn’t feel right to pick that value?
You’d benefit from reading some of the blog post & explanatory material that TR has issued for this release so that you understand a little more about how the system and new features work.
Regarding your first question, TR AI will attempt to give you workouts at certain wattages; if working with a higher FTP number, these wattages will necessarily correspond to lower workout levels in order to deliver those wattage levels.
The “black boxes” you cannot (currently!) remove is a visual representation on the Calendar of the 28 day Simulation Window, which is a key component of this TR AI release.
The system “keeps altering [your] chosen workouts” because this is a fundamental feature of this TR AI release, whereby a new type of workout (“AI Workout”) exists in which the AI system now chooses the specific workout for you to do based on its model of your capabilities, fatigue levels and its knowledge of “what works”. You can override this behaviour on a workout-by-workout basis should you wish to, but normally you won’t wish to.
I think you need to spend a little bit of time learning about this new release, and this will serve you very well in making better use of it and asking questions about it. The good news is that generally you can just do as you wrote, “just doing what pops up on the calendar”, letting the AI system do its thing, but it’s still beneficial to understand what’s happening!