Not a good idea, IMO. The day to day variation we see in FTP is “noise” in the truest sense of data analysis and is influenced by all the typical factors we know.
Unless something drastic like a sickness, injury or other reason leading to a lower FTP happens (that TR already gives at least some level of user control via related Annotations), there is minimal chance the AI FTP will give a big difference from one day to another. Any variation people have from one day to the next is unlikely to be large enough to shift training zones in an impactful way.
The lines drawn between zones may appear sharp and it’s generally necessary to do so for a system like this, but the the reality is that our physiology is not nearly so discrete or influenced to such a minor degree when flirting with those boarders.
Most notably, there are two things TR has said (largely via Nate in podcasts and the forum) that counter this idea to a large degree:
Nate has mentioned issues with people looking at AIFTPD too often. They feel that leads to issues and have made restrictions to how often and when people can access the tool.
Somewhat counter to that, but more aimed at the future, Nate also said that their long term goal is to have AIFTPD be largely automatic in a way that it alerts users to a notable FTP change if/when that happens. This would apply to increases and decreases, and is aimed at more deliberately catching “real” changes in FTP vs some predefined schedule that is currently in place via Training Plans and Plan Builder.
With all that in mind, I just don’t see a benefit at all and more likely problems with the suggestion. I think the current & pending direction TR has outlined makes the most sense.
I personally changed my ftp too much when the feature came out. Yup I was one of those guys checking daily, even 5 watts felt too much and failed more workouts then ever before. It set me back IMO, I think it’s better to get your levels up and then change FTP if it’s the start of a new block. In the end I kept my ftp at a realistic figure and built my TTE.
Definitely. It gave me some (limited) insight into how the AI models fitness across a block of successful workouts with consistent evaluations. Mostly it was intellectually interesting.
Out of curiosity today, I did an AI FTP, followed by a ramp test. AI FTP 290, Ramp 270. Goldencheetah CP (I know, not the same thing…) for last month 280. I’ve been doing TrainNow along with outside rides and not on a plan, with FTP manually set at 285, and most workouts have been doable (exception: high power VO2 max, which is probably why Ramp underestimates for me)
Anyway, none of this is criticism, just feedback for TR folks. Maybe 290 would work if I were following a plan, who knows, but at least 290 is closer to what I feel my “real” FTP is than 270, so it’s an improvement over the ramp test (though unsurprisingly for me, I always felt the old 20 minute was better)
@IvyAudrain I’d like to add another voice to the #feature-request for some sort of clearer visualisation of when AI-FTP will be available. Some sort of countdown for the 14 (soon to be 28 days).
I have a Ramp Test today on the schedule as the start of Build. No AI FTP is appearing though I’m reasonably confident that I last used it more than two weeks ago to see what it would suggest as I restarted post-Covid.
Joining the chorus, I also have a ramp scheduled tomorrow and no AI detect button. My last AI detect was 6/2 so it has been longer than 14 days. Maybe they made the change to 28 days on backend?
Interestingly I’m not entirely sure why AT put the previous ramp test on my calendar as it was mid-block. One day it just proposed an adaptation to replace a VO2 wo with a ramp. I could have done something to precipitate that but I can’t think of what.
Good thought but didn’t work for me. Changing date to today did make it finally show for me but interestingly I swear I did that earlier and it didn’t work. Almost seems like the combination of setting the time to 12am AND today made the detect button finally show, which makes it seem like the schedule date had to be in the past instead of in the next 24. If that is true then that may be an inadvertent bug. I only share for the purposes of potential beta feedback, not to complain.
While I like many are eager to see what the result will be, my primary use case to do the detect the day before the scheduled ramp is because I know it will then replace the ramp with a scheduled workout. I intend to do that workout outdoors and want to get it sync’d with Garmin tonight so it’s on my head unit and ready to go so I’m not mucking with it at 5am tomorrow.
I set the date for mine to yesterday and it finally let me use the AI-FTPD. Weird and not sure what is up there. Still having a weird syncing issue with Garmin but might (most likely) be completely unrelated.
Hello,
I have “a question”:
Does it make a difference if I use AI-FTPD after completing my last hard workout before the scheduled ramp test (only 9 days of endurance easy week in between)? How about 2 days before the scheduled ramp test (only 48h of no training at all in between)?
Does this give me a lower ftp or would it be the same as if I only used it on the exact day of the scheduled ramp test?
Quick update: we just deployed some changes that will alter the minimum required data from 12 workouts to 10 workouts.
This means that in order to use AI FTP Detection, you will only need to complete 10 indoor TR workouts. This change will affect all athletes currently in Early Access as well as any who enable this feature from today forward.
Anyone else curious when TR activates AI FTP to update your FTP?
In my case, it’s been over 4 weeks since my last FTP update. Early Access is enabled. I have had combination of 15 indoor/outdoor workouts with power data.
Is there another condition that needs to be met? Yes. From reading posts above, needed to schedule Ramp Test.
, but it’s a minor inconvenience.