Yes, how this worked changed during the beta period, so if you were a beta tester you could have seen predictions without accepting the initial detection.
I think the best way to see the new TrAIner Road ftp especially if it’s lower than previously is like the difference between Fahrenheit and Celsius. 25C isn’t lower than 77f, they’re just different numbers forth same temp. As the new ftp is a different number for the same level of fitness. The system runs off different scales
My FTP also dropped 20 watts (258 to 238) after a few months of work over the winter.
Few thoughts on this -
The new FTP actually feels correct. Previously, I wasn’t able to complete a threashold workout beyond level 3.5 without serious mental anguish, whilst SST was limited to shorter durations in order to complete (probably more like threashold workouts).
I think there was a vast library of workouts which I was constantly not getting access to by increasing my ftp and lowering my PLs. I actually think these workouts I was missing were the key to me getting faster and stronger. One solution would have been to lower my ftp and extend progression levels, which is what it feels the new system has done.
Fingers crossed, training at this new level and pushing TTE through higher PLs will be the thing that pushes me forward. Previously, there would be a peak, followed by a collapse. This new AI feels more like a steady raise of aerobic systems.
My ftp dropped 40 watts. I’ll continue to use TR because of it’s breadth of workouts, but I would take the AI with a grain, er, a truckload of salt.
The old AIFTP predicted the outcome of a ramp test.
People can over-perform on ramp tests. Ramp test results don’t always give the best FTP to train at.
What makes you think the result of a ramp test gives you either:
An accurate estimate of your actual FTP; or
A good number to use to prescribe workouts?
Thought experiment: Do you trust it enough to do an open-ended threshold interval at that number and see if you can hold it for 40 minutes or more? I know I wouldn’t, especially if I’d done 3 weeks of 30/30s to prep for the test.
Pretty spot on confirmation of the old model. Which then would lead one to think that they know what they are doing with the new model.
Been confused since trying the beta, as did sweet spot 3x20 at the new ftp few weeks prior, and 4x15 a week or so later. Posted the screens up in the beta thread. Subscription run out thinking of rejoining but I’m not sure. My ftp always felt about right (TR AI FTP figures), and if I do a 10 mile TT at this new ftp I’ll be going 20 watts slower. Guess I ignore the 105% I follow and go 110%. And ignore training zone colours from now on the Wahoo, as the zones are different.
Can’t get my head round it.
exactly! OP is looking at the glass half empty. Accept the new FTP, and watch your PL’s soar. trust me, it will improve your cycling, and be alot of fun.
In addition, I might be wrong, but oftentimes higher level Tempo workouts are at the same power percentages than lower level Sweet Spot workouts. Higher Level Sweet Spot workouts are similar to Lower level Threshold workouts.
So take the lower FTP, allow PL’s to climb. your workouts will be the same intensities that you are used to, but will be tagged a different power zone.
As always, hope this makes sense.
I see what you’re saying. I just can’t accept that my ftp is lower than it was before 6 years of structured training and racing competitively. ![]()
But does it matter? If TR made you faster before, it will make you faster again. If you like the Ramp Test as a baseline test, keep doing it and you should see it increase.
dropping someone’s ftp inside of 300 watts deserves jail time without parole. hard labor, and broth for food… once a day.
DAAAAYUUUM!!
Simply because that’s how we have always done it. Maybe I’m just a stubborn old man who resists change. Lol. And just for the record, I’m not trying to say it’s wrong or that the new version isn’t better. Just trying to understand.
TrainerRoad AI just dropped, and I was quite excited. However, the first thing it did was recalculate my FTP, and whilst I was expecting 234-238w, possibly even 240w, it estimated my FTP at 218w!!
WTAF
Give it a try. For reference, mine went from 294 to 272 BUT i can tell you my workouts felt A LOT better AND have been more encouraged by my recent training. The new probability scores have been better at setting expectations of how the efforts should feel.
While it’s a bit of an ego hit/demoralizing to see FTP go down (I get it), the new system is putting you in a better zone to progress.
My FTP was 216 after last AI update before the new launch. After the launch it gave me 203. While I was disappointed I was planned on giving it a go. Now today it predicts after 18 workouts 182.
Now what AI? I’ve been a TR user for 12 years. Im 56 and I realize the age and the inconsistency of my recent training. But this seems off.
If your training approach isn’t set to Balanced, change it to Balanced and see what happens to your prediction.
If that doesn’t fix things, maybe post your last 4 weeks and next 4 weeks from your calendar.
The old AIFTP predicted the outcome of a ramp test.
Hang on, lets not chnage history, this forum for the last 1 or 2 years has been telling us that, ramp tests are rubbish and that AIFTP produced better results, the TR video about sites it being 45% less likely to underestime your FTP, I don’t think anybody in this forum has said this beofre
Nate mentioned it in the beta thread - first I’d heard of it either to be fair.
This.
I’ve apparently gone from:
229w AI FTP 28/12/25
222w new AI FTP 22/01/26
194w predicted FTP 23/01/26 ![]()
That’s lower than I was 2 years ago. All the work apparently just gone overnight.
I just don’t trust trainer road anymore.
Although I suppose I should actually thank the team, as this has given me the push I needed to cancel my subscription and spend my money with European companies instead.
