I just got my latest AIFTP estimate today and it’s gone a bit off the deep end (on the high side) for me. AIFTP has been running ~15w high for me since the first estimate (compared to how the previous version estimated FTP). But today’s estimate is at least 20-25w high.
My training program this year is very similar to last year with just a slight decrease in my ramp rate and overall volume. And my fitness is a tick lower as well, not quite able to hit the workouts I did at the same time last year. But I am into the part of my plan where I’m pushing a lot of volume/TSS and doing a lot of long/hard efforts. I’ve been mostly picking my own workouts and managing my own build/load this year and AIFTP seems to love what I’m doing. My FTP bumps keep exceeding what TR projected under the prescribed plan.
So, with slightly lower fitness than I had at this time last year (when my FTP peaked matched an all time high of 314), AIFTP is currently estimating my FTP at 335 with a prediction of 340 in 4 weeks. Using the old AIFTP approach (or any reasonable physiological definition) for FTP, there is no way my FTP is over 315 and I’d say it’s ~310 based on recent efforts. Just last week, TR told me that an O/U workout with 2’ overs at 340 was “not recommended” and would be too intense. If I truly had an FTP of 335, an O/U workout with overs at 340 should be super easy. So, the TR system knows I don’t have a physiological FTP of 335 or it wouldn’t have steered me away from that workout. I guess this all comes back to trying to normalize around workout level 3, but it’s just getting further away from any reasonable physiological FTP definition for me.
I have to believe it’s the volume that is swaying AIFTP. I’m not doing anything spectacular in my interval workouts and have even failed a few workouts recently (trying to stay on pace with what I did last year). And for me, that just highlights one of the biggest challenges with the new AIFTP number for me. The system is telling me my FTP is over 20w higher than what the system told me at this time last year. But I’m objectively less fit based on my inability to hit the same workouts. The system is still doing a nice job of estimating the difficulty of workouts, I’m enjoying that feature in the new system. But I will continue to set my FTP manually unless the TR AIFTP number eventually aligns somewhat with a physiological value.
I did accept the new number for a moment just to view the fictitious 4.5w/KG on my home page. I’m nowhere close to that, but it’s still cool to see and dream.
I’m in the same boat. I’m progressing well, but my FTP number is probably 20 watts higher than any other way to define FTP including rampt test, 20 minute test, 1 hour test, and the original definition of FTP (the highest power output a cyclist can maintain in a quasi-steady state for approximately one hour without fatiguing). All of those tests and definitions line up relatively close to an FTP that is about 20w lower than what TR says my “FTP” is.
TR basically has the stance that their FTP number functions in isolation inside the TR ecosystem. If it is so seemingly arbitrary, why don’t they just line it up with the traditional FTP number? The AIFTP is really good at prescribing workouts of the appropriate difficulty and my progression is going well, I just don’t get why the FTP number is inflated.
I’ve just done a quick review of your recent training history as well as this time last year, and with our current AIFTP model, we would have given you an FTP of 327 on April 11th, 2025.
Back in March of 2025, you were doing mid to high-level 6 Threshold workouts with an FTP of 310. Last week, you did a level 6.6 with the same FTP, so I’d say that your fitness is probably comparable to what it was last year (at least in some aspects).
Looking at the model today, I’m seeing an FTP of 324 (just slightly below that 327 from last April), so I’m not sure where that 335 came from. Did you view your FTP before, in between, or after the two activities you did yesterday?
I guess that’s kind of my point. Last year, AIFTP gave me an FTP of 310 on March 14th and this year it was 327 on about the same day. We can debate how close my fitness is tracking this year vs. last, but at best it’s similar. I can point to several workouts I’ve failed this year that I hit at the same time last year.
I don’t know anything about models, I just know what the system is showing me. When I logged in yesterday (before riding), the “new AIFTP” dialog popped up to give me a new number. I accepted it (just curious what my w/kg would be at that number) and immediately deleted the record. But it’s still on my career page waiting to be accepted.
And if we want to compare that 335 to what AIFTP estimated last year (314) - again with similar or slightly higher fitness this year, it’s a difference of 21 watts.
I know it’s probably beating a dead horse, but this makes year to year comparisons between TR AIFTP numbers a bit pointless.
I think this is what has a lot of people struggling. There is always going to be healthy debate about how FTP is defined, but I still haven’t seen a clear statement from TR saying what performance measure AIFTP might approximate. I think I’ve seen some comments saying it still should be a decent performance proxy and aligns with traditional concepts, but then there is the “level 3 normalization” as well. It was a big new release just a couple months ago, it’s reasonably to expect things not to line up for everyone at the time of launch. But it seems that it’s all working as intended from the TR point of view, so the people who are outliers kind of feel like “it is what it is, deal with it”. Personally, I still find the system very good overall, but AIFTP isn’t something that works for me in it’s current form.
What I was saying early in my message was that you’re doing very similar workouts right now to what you were doing at this time last year. It’s hard to say exactly where you were at last year from a glance, because the three threshold workouts leading up to that 314 detection all had wildly different RPE surveys (moderate to max effort).
The model I’m referring to is the current AI FTP Detection. Rather than comparing your detections now to that 314 number, which is sort of like comparing apples to oranges, I thought sharing the 327 number would help and be more like apples to apples.
Knowing when you got that last FTP Detection helps.
In terms of a “traditional” FTP number, the biggest issue, in my opinion, is figuring out which method is the most “traditional” and if that’s the best way to go moving forward. If we don’t think that method will bring the best results, we’re happy to step up and make changes and lead with what we believe in, rather than use something traditional for tradition’s sake.
The most common definition of FTP that I’m aware of is 1-hour power, but I really don’t think that’s the best way to define FTP, nor the best benchmark for training for most athletes. I’d be curious to know what the top-level coaches use for FTP, but I’d bet it’s not hour power. We can’t use lactate for obvious reasons, and RPE isn’t the best option, but with the dataset we have, we think that using a level 3 threshold workout as a benchmark/starting point for our athletes covers the vast majority in the best way.
Any athlete having issues with their training since the launch of TR AI should definitely reach out to us, as we’d like to look at their case to see if there’s anything we can do to help or improve on our end to make things better for that athlete and any others that might be experiencing something similar.
In your case, I’ll definitely start a discussion with the team. Please let me know what type of issues you’re having with your training. If you’d rather discuss this via DM or support ticket, we can go that route. It’s totally up to you!
Agree that it’s a complicated issue with all the different ways to define/measure FTP. And not looking to devolve into that discussion. If TR’s position is that it’s strictly a number to normalize training within the TR system, then I think it would be good if that was clearly stated. I believe it’s still being positioned in places by TR as a physiological performance metric/benchmark. And if that’s the case, it seems like it should align with the “traditional” concept of FTP. Lots of wiggle room there, but all of them point to FTP being a threshold for continuous aerobic power.
And I’m sure the new AIFTP is aligning great for the vast majority of athletes, I just know the old AIFTP aligned very well for me and the new one isn’t close. I got an AIFTP number of 335 at 8 AM yesterday (which I ignored) and just happened to have a vo2 max workout with 5’ intervals at 335 an hour later. With 5 minutes of rest between each interval. If my FTP was anywhere near 335 (or even 224), that workout would have been a really, really easy workout. But it was not and I only made it through 6 of the 8 before falling apart and skipping the last 2. It was a good vo2max workout, not a threshold workout. Having a properly set FTP keeps the workouts aligned with their purpose (as described in the workout text).
The AIFTP thing isn’t causing any show-stopper issues with my training, I just have to set my FTP manually. It hasn’t been a burden because I’m mostly following my approach/workouts from past years and my performance in workouts is within a handful of watts to what I was doing at the same time last year. But yeah, I’d love for the new AIFTP to more closely align with the old AIFTP (which aligned very well with threshold and O/U workouts, etc. for me). Besides making workouts appropriate, being off by 15-20 watts between the old and new approaches makes it hard to measure performance historically. Maybe I’m the only one who is constantly looking back at my history to see how I performed based on what training/workouts I’m doing, but I find that type of analysis very beneficial. I’m happy to engage with support to answer any questions about how I use FTP in the system and how I typically validate that the number is close. Thanks for taking the time to respond.
Look at the power curves you did last year and compare them to this year. Eddie has said the current AIFTP method used on your rides last year results in a nearly identical AIFTP from this year. An easy way to verify this is to just look at the graphs.
Imagine you’ve been using a bathroom weight scale for 20 years. It’s one of those old analog ones that has physical components. You’ve used it for a long time, so you’re very comfortable with how it works, how your weight fluctuates throughout the week and during the year. But it has probably worn down over the years, it’s rusting, and it’s not very precise.
Then your wife buys a new bathroom scale, its digital, and it has a properly calibrated load cell. You step on the scale for the first time and you furrow your eyebrows. “…there is no way I am 5 kg lighter”. You probably aren’t lighter, the previous scale was likely wrong (but still useful since it was the only place you weighed yourself). However, it’s also why you can’t compare your weight on the old scale to the new scale. The same weight on two different scales will read differently.
That’s what you’re doing comparing last year’s FTP from the old FTP model to this one.
The system is still doing a nice job of estimating the difficulty of workouts,
Then the new bathroom scale is almost assuredly working. All that really matters is assigning you good workouts.
If you use the FTP number for pacing during races or something, then maybe the TR AIFTP doesn’t apply (or maybe it does). But I’d suggest you try it and see for yourself.
I have and decided that setting my FTP manually works better for me.
All of this has been beaten to death in other threads. Bottom line - using the new AIFTP (which doesn’t align with any meaningful physiological wattage) causes more pain than benefit for me. System works better for me with a manually set FTP, I just lose AIFTP detection. It’s a good trade off for me, YMMV.
I’ve been giving it a shot using the AIFTP and following the plan, but ended up getting the old over/over threshold workouts that are just long VO2 sessions in reality, and my SS work ends up closer to threshold intervals that don’t extend out that long. So I think I’m going to do the same, and manually override the FTP lower so that I can get back to doing longer SS work and keep that durability high.
I was hopeful and wanted to give it a shot for a few months, but I’m losing confidence in the way it is trying to progress me unfortunately.
When you set your FTP manually, does TR still adapt the workouts automatically?
Did a ramp test right after AI detection and it was about 10w lower. Looked back and had similar delta in the past too. That explains why the endurance rides always felt like tempo rides and threshold workouts had my HR into VO2. They all were rated appropriately as low tempo was easy and high threshold/vo2 was hard.
I’ll try manually adjusting my ftp after I complete my block.
Yes, TR will still adapt your workouts, you may find that it will give you longer interval lengths as a result though,if you lower your FTP, which results in more TiZ, but personally I don’t see that as a bad thing, other opinions are available.
Yep, and this has been the case for me and a few other vocal folks, including the OP. TR, at least from my reading and podcast listening, has not addressed this issue. I get that AIFTP somehow relies on your ability to complete a level 3 threshold session, but what hasn’t been addressed is using that number for some people, your z2 pushes into Z3, SS pushes into thresh, and o/u’s really become longer vo2 session. Jonathan did one of those ‘reviews’ of my training, and he basically said, you’re good man, keep it up. Before I cancelled my subscription last week, my AIFTP was into my vo2max, and a good 15-20 watts over my threshold. I could somehow still complete a level 3 threshold session though! Just becasue I can survive an o/u workout does not mean it works for my training in my experience.
I’m fine with a variable definition of FTP, and if it is being used mainly as an anchor for training. But if an over under workout isn’t giving you any level of recovery during the under, then I would say it isn’t really doing what it is supposed to. The FTP being used should be somewhere in the middle of those workouts, and if not I feel it isn’t really working.
Before AIFTP I was doing a lot of customizing of my training anyway, so I’ll just go back to that. But I can definitely see if I was earlier in my training journey I’d just be killing myself trying to complete these workouts and getting frustrated/burnt out.
Agree. I’m hoping is they release an improvement in the future where the user can choose what level of threshold the AIFTP anchors - something like level 5.5 or so makes more sense to me personally. As you said, the unders gotta feel like unders for that type of workout to make sense. Same with the other zones.
What I did, which effectively is the same thing, is manually lower my FTP - you can basically dial it in so you do get the say a level 5 or 6 threshold workout. It would just be nice if you comes out of the box like that and you get the benefits of seeing your predicted FTP (you don’t get that when you go manual).
Same issues here but once I ignored the “inflated” FTP, I soon realized the prescribed workouts were all spot on. The new system works better for me. I never tell people the FTP that TR says because I would get the bs look.
Interesting observation for me - my AIFTP prediction was trending pretty darn close after being 5-15W high earlier in the winter to the point I was ready to roll with it. But then on the day of the new detection, it jumped +5 from the prediction that I’d had just the day before. Still relatively close, but now definitely 5-10 high for me (maybe actually 10-12) so still running with the manually set FTP.
Similarly, I’m doing threshold work pushing higher/harder athlete levels there and layering in some of my longer rides starting the process of working volume up, but I’m still earlier in my volume ramp-up and maybe 30 points below where my CTL will peak later this summer.
My experience with my coach, is that the WKO mFTP works very well when fed the right longer steady state efforts, but it’s way, way off at certain times when you’re not giving it that data. And in general what we’ve done a lot of work on isn’t a specific test, isn’t a specific metric - it’s spending a lot of work feeling that tipping point where you can warm up and then pretty quickly settle in to riding at FTP. Absolutely something that you can’t give to a newbie though.
For example - I’ve done variations of the Kolie Moore Tests, but I don’t always do them religiously (but when I have WKO number is pretty dead on.) Honestly, a standard 2x20 @ 100% works fine though. I’ve gotten to the point where I can be at my FTP for 5 minutes (fine), 10 minutes (fine), 15 -20 minutes - realize I’m just a little high and starting to spike / blow up a little. Then - go into the second interval, drop the wattage 5W, and hold it straight through and see what happens. Ends up being a hard productive threshold workout, while working on “feeling” FTP and working in on the right number.
For me - it’s a combination of RPE and watching my heart rate. Does it ramp and start to stabilize around 90% of max (my threshold HR) or does it start bumping to 91%, 92%, and so on while RPE starts to spike? Or, am I not bumping up against that threshold HR and running a little conservative? (I had a spin bike workout earlier this week where I was able to do a shorter but productive threshold workout with no power using this strategy and I’m sure I was pretty close to FTP, maybe just a bit low on purpose)
Works as you get more experienced, but not at all for people that need more of a test or a metric.
One more thing - as someone gets closer to their FTP Ceiling, they get to the point where all they can do is extend duration at threshold. That’s when the Level 3 goes out the window. But at the same time, my last detection did not leave me in the 3’s when I accepted it - it was in the 5’s. The current one when accepted puts me at a 4.6 as of this morning.
You’re correct. I lowered my ftp based on my ramp test and the workouts were extended or more time in zone. I wish tr didn’t assume by me lowering the ftp that I can do a higher workout level. Did the Eichorn 2x20 @94 ftp and barely made it to the end. This is coming from completing a specialty phase with no sweet spot workouts for a long time. If only they ramped me up slowly by starting with 4x10 or something.
