love this! Can relate.
I started with the 20ā test protocol, & for the first few AIFTP detections I was able to significantly outperform it with a 20ā test a couple of days later. Since then Iād say AIFTP has got very accurate: trying to beat it thereafter has been a case of putting myself through the wringer for one measly watt.
4 Likes
Can you explain a bit more? I also see vo2max training comes much easier for me than threshold and endurance. I recently did 10.4 level tempo indoors I and progress faster with vo2max, but threshold is always painful. I wonder what should i do about it.
Iāll try. TR prescribes VO2 Max work by power as a percentage of FTP, say 110% of FTP.
For a rider with low anaerobic capacity, this is probably a very good intensity at which to train their top end.
For a rider with high anaerobic capacity, itās likely to be too low an intensity. It just wonāt drive their heart and lungs hard enough.
If threshold is really hard on your legs, then you likely just need more aerobic work. Long, easy endurance rides and a sprinkling of sweetspot is likely a good approach. And practice at telling your legs to shut up.
If threshold is making your breathing go ragged and your heartrate rocket ā¦ you likely either need much better cooling (if youāre indoors) or you have a bad FTP number (too high)
5 Likes
Mine was fairly accurate, AI @ 281w and my actual test was 285w
2 Likes
AI FTP appears to provide a sustainable level of training allowing me to maintain, improve, and recover. Could I hold the AI FTP power for an hour? Hell no. Does that mean its incorrect? I dont think so, as its only relevant to me in the context of TR.
I am not a pro, I only have time for low volume and sitting at around 3.3W/kg at 44 on low volume, as well as 2x gymnastics and a 3-4 hour MTB ride on Sundays is pretty good I think!
5 Likes
Those are some mega gains - not surprised AI FTP took a while to catch up as its model isnāt just based on best efforts like some others - but also your history as an athlete.
Also. are you 100% sure on all your PM measurments? Could be that the original measurments were a bit on the low side?
1 Like
Welcome back to TR! 
Nice job with the training consistency since getting back into it after taking a few weeks off. 
I think this instance seems to be a bit of an edge case considering your long training history and the fact that you were coming back from some time off.
Itās normal, of course, for your FTP to drop after taking 6 weeks away from training. Now that youāve gotten back into it, though, since youāre an experienced athlete, I bet your fitness has been coming back FAST ā possibly faster than AI FTP Detection was able to predict.
The model works based on the data it has, so from AI FTP Detectionās perspective, it sees that youāve taken some time off, done a Ramp Test, and started out at an FTP of 220W. That first jump in your AI FTP Detection from 220W to 227W is, funnily enough, about a 3% gain ā but, based on your recent training consistency combined with your previous endurance training experience, it makes sense to me that your FTP has gone up more quickly than the model was able to predict.
If youāre ever in doubt about your FTP, I think doing a fitness assessment is a good move so you can feel more confident about where youāre at. As a bonus, doing so also gives AI FTP Detection more data, which means your future detections will be more accurate.
That said, now that youāve gone through a few training phases and are likely starting to see your FTP āsettle inā a bit, weād encourage you to see how your current threshold power feels as you get into sessions with Workout Levels that get into the 4.0+ range. That tends to be about the turning point where workouts start to get more āsubstantialā and demanding on your fitness. Lower-level workouts can be a little easier to ābrute forceā your way through if your FTP is too high/low, but that becomes more and more difficult to do as interval durations and intensities increase (this is what @alexgold123 described above).
I mention that because it is possible to āoverachieveā on a fitness assessment if youāre feeling particularly good on a given day ā which can then make the rest of your training more difficult than it might ought to be. Thatās another reason why we advocate so strongly for using AI FTP Detection once it has a solid set of data to work from ā it looks at all of your recent training to dial in your FTP, which means youāre less likely to get a skewed FTP value.
Hope that helps clear things up ā keep us posted on what your future AI FTP Detections look like, weāre curious to know now that youāve been back on the bike for a while!
13 Likes
I donāt know if its a TP Virtual (previously IndieVelo) thing. I noticed it for the first time last night but Iām not a signed up TP member. Iāve just activated the TP Virtual trial.
I use my H3 as the power source and I have compared it to my Power2max on my roadbike, they are pretty close. To add some contextā¦my FTP was 285 last time I tested late spring of 2024 so it jumping up quickly after 8 weeks of training is expected. I also used to be between 300-315 6-7 years ago when I was racing both road and MTB. Like Zach mentions below, with a long history of training it comes back quickly and AI might not catch it.
1 Like
Itās probably just a personal hang-up, but I often find myself thinking a lot of things called AI should more accurately be described as Data Driven Decision Making. For some reason, Iām more willing to accept errors when itās called the latter.
4 Likes
In general, how I feel about the accuracy is I like it when it goes up, I think itās off when it goes downā¦
In all seriousness, it seems to work for me and goes up and down based on my consistency with structured training. tbh, Iād got to the stage I was adjusting on feel and verifying with Lamarck pre-AI FTP as I just sucked at the Ramp Test, and you had to have some idea for 20 minutes or KM Protocol.
Is it my actual FTP - not sure, but it appears to set my workouts at the appropriate level if I go in properly rested.
10 Likes
I would get a message from TP that I have a new threshold. I forget if it was via text or email, but Iāve had it happen a couple of times. It would not change my zones however, I would have to do that. But Iām no longer actively using TP these days but still have my account active.
1 Like
TP will send you an email about threshold increases after you trigger it via some sort of sustained power of X time.
Also intervals
Iāve done the KM tests and had both send messages as well basically doing the math so I donāt have to. So it isnāt āAIā just some sort of preset triggers they both have.
Its in settings and notification, it is as old as gods dog, been there forever.
You can get a notification or it applied automatically.
Note: Nothing to do with TPV, its always been there.
1 Like
Nailed it.
Iām naturally more anaerobic so V02 workouts arenāt really challenging until PL 6 or so.
A threshold workout @ PL 3.5 or so would feel equally as challenging.
4 Likes
Huhā¦I donāt think I have ever gotten a notification from TP re: a suggested FTP change (and I just checked and I do have it marked.)
Guess that tells me what I need to know about its value. 
TPV has a CP chart and gives you your FTP, Critictal Power and Wā Prime, at the end of a ride it tells you any all time PBs on your power duration curve PDC as well as 30 day bests
Do you maintain your threshold in TP though? Or is it set a glory days value 
Iāll get a notification 2 - 4 times a season, slightly less often than Garmin
1 Like
Absolutely. Every time TR adjusts my FTP, I adjust it across all platformsā¦TR, Zwift, TP, Garmin, etc.
4 Likes
Fair enough then.
I find it only gets triggered if you get a recent best in the 20-25 minute region by some margin and/or in the 5 minute region.
I used to think it was 95% of 20 minute power based, and maybe it is but Iāve also had 5 minuters trigger it so started to think CP model, but its more likely that the 5 minute ride/session has a good 20 minute peak in it. So who knows.
1 Like