You can just manually type in your FTP and youâll get levels based on custom workouts (soon) and outside unstructured rides (in the future).
It does not yet. But weâve got some ideas on making the different systems flexible. We still need to get AT out to everyone before we add these changes.
Just so Iâm clear: it wonât make a difference to AT how you generate your FTP, correct? So Ramp Test, 8min, 20min and manually entered will be handled the same way by AT?
Just another question, small one.
Will the rollout of AT come with a comprehensive manual / wiki?
I just love to read but I also think being able to understand the nuts and bolts of the approach in detail will be much appreciated.
WaitâŚI donât like this. I want it to hurt more for everyone else in the group ride.
I wonder if that can apply to individual workouts too. if you measure hrv during warm up and it indicates youâre fresh it could raise the intensity a bit and if it says your worn out it could lower it a bit. Though hrv is generally better at trends and may not be sensitive enough day to day to do this
Super excited to be able to use the Group Workout function (in any shape). Iâm curious about the hardware requirements for it on mobile, is it quite demanding, or is it also done server end and just requires a strong internet connection?
Cool! An exciting first step 
Just suffered through a 1x120min@90% (final workout in WKO extensive aerobic phase). Guess thatâs >= 10 in SS progression.
So if I now start into SSB2 for exampleâŚwhat does AT do/âthinkâ? Maybe it would recommend FTP raise (IIRC nor there yet.). What if I stay on that FTPâŚwould I often get the hardest SS workout of the catalog recommended for SS days?
I really like this idea. I think there are a lot of people who have settled on a custom percent, maybe 72-73% after a year or two experience using the ramp test and seeing how they perform on subsequent workouts. I think the AT will ameliorate this greatly, but if I already know things progress better with the 0.72 ratio, it would be nice to be able to automatically use that starting point.
That said, we can always manually update anyway, which is what people do now.
Does adaptive training differentiate between a completed workout that just kills you versus a completed workout that was easy? Does it use heart rate data to know what was hard and what was easy or do we have to tell it? Clearly it can tell if you failed the workout or not, but there is huge variability within a âsuccessfulâ workout in regards to the level of effort that successful workout required.
Thatâs what the RPE survey at the end is for.
Correct.
No, itâs pretty easy to use. You just ride and accept the adaptations.
Just turn it down 
I suspect if we could ever tell during the warm up if you were worn, wed just end the workout or make it easy.
Internet is like 1mpbs needed, so not much.
They did a POC on mobile and I think it wonât take much. Iâm not sure though.
@Nate_Pearson Might SMO2 and/or THb data from a Moxy sensor provide some useful insight to how prepared the muscles are to workâŚ? Iâve just started down this rabbit hole with a Moxy sensor, and think there is potentially some very insightful data that could be collected/displayed right along with power to help a user identify if they are âover-reachingâ. I.e. SMO2 that is continuing to negative trend during a longer, or later interval, could provide real-time âinformationâ to evaluate when trying to decide what to do with a current workout.
Check out this workout I did the other night with the Moxy. Intervals.icu
Had this been a 4x15, the tail end of the 3rd interval might have been a good indication to âpull the plugâ, as the last ~3 minutes showed a declining SMO2. I havenât spent enough time with the sensor yet to know if that would necessarily be the case, but presumably ML could pick up on those types of trends if they correlated with future failure? Or maybe something from the warmup would even indicate; or maybe the SMO2 response of priming intervals after the warmupâlots of thing could potentially give good indications of what to do with the workout.
Realizing that link might not work, so including a pic of the workout data
listening to adaptive training and pretty suer my coach wonât like this much⌠hehehe at the moment I use a coach, that give me workouts, I code them via workout creator and then do them. this sort of lends to me not doing thatâŚ
question, Jonathan eludes to inputting the âfeelingâ of a workout after the workout so that the system can adjust as/when needed, I much be blind⌠known to happen but where do we do this, so that the algorithms know how to adjust the IF of workouts being prescribed.
G
I hope TR supports the data. At least to just record it and feed the model if you donât want to use up develoment resources to add to the UI