2nd cycle done for me
17/2 prediction 387w
17/03 ai prediction of 400 w
Today 17/03 prediction 386 w , down from a month by 1 watt, wasn’t surprised, several failed workouts again, and training continues to be a grind. Compared to previous years where I’ve seen progression in workouts which makes it an achievement, just seems to be slow decay this year.
next prediction is 399 watt…
14 Feb 214 watts prediction 224
14 Mar 213 Ai FTP prediction 221
I failed a threshold workout and missed a couple of endurance workouts because of outside distractions.
Threshold seems to be what I struggle with,I can do sweetspot and VO@ max but threshold kills me. Even though my FTP didn’t really move I feel like I’m a stronger rider.
Further update after another month…
I’ve made great progress sticking with the AIFTP detected FTP number…. I still seem to be growing into it which is a great feeling.
Last cycle started at 333W predicted 349w actual ended 336w
This cycle started at 336w prediction is 351w with 17 days to go.
There is absolutely nothing in my training history which would indicate that I am capable of getting to 351w - but I’m confident that I’ll get another decent bump - my money will be something like 338-340w at the end of the block?
I did a level 4.5 threshold workout yesterday and it just felt “hard” - I feel like my training is in a great place - good work TR 
I’m going to throw a spanner in the works this weekend though - I’m on a recovery week but have entered a half marathon running race on Sunday that the AI doesn’t currently know about and I haven’t trained for 
Initial AI FTP Prediction for today: 406 W
AI FTP Prediction for today after optimization via Check Volume: 412 W
AI FTP Prediction fluctuation over the last 4 weeks: 408 - 413 W
AI FTP Detection today: 413 W (based on yesterday’s prediction after my workout)
I decided to push the actual detection to end of the week, because next week is a rest week and it feels a bit cleaner this way.
new AI FTP Prediction for 4/14: 425 - 427 W (depending on if I do today’s 30 min endurance ride as an actual TR workout or as a solo ride 
)
My prediction has plummeted and reversed last month’s gains 
I had a mechanical issue in the second week, then skipped workouts in the third and failed a threshold due to illness. No way to recoup in the fourth week!
Frustrating but fair.
Second cycle for me too. 17/03 real FTP detection was 211 downgraded from 243. I do not follow a TR plan as i said. I do tempo workouts as they suit me for now. Just ad-hoc tempo workouts to the calendar and AI does picks “spot on workouts”. My consistency was meh, close relative passed away and i did not have the time to train hard. I think that the new ftp is about right. Nevertheless i am feeling really good on bike. Slowly getting to pre fatherhood form, still 30% off it. I am at cruising the local 4k 7% hill cant wait for snow to melt and try the 18k mountain road with 1k elevation. Next prediction 17 April is 226. Still below the very first detection of 238. TR AI for now objectively tracks my “FTP-non FTp thing”. I think i have plateaued from only doing tempo its time to move to Sweet spot- balanced/ conservative approach.
Achieved 276 so a gain of 16w to land at exactly 4w/kg!
Did 90%of the work inside on the trainer with 1 key session dropped for an unstructured mtb ride in the hills.
New prediction is 287w so another 11w!
System seems to be working amazing for me. I’m getting gains each month that took 6 - 10 months to get the last two seasons on TR.
december 15, restarted TR after using a coach for the past year (ill likely resume with him but he has some life things going on). 217 which felt insanely easy as my previous setting on training peaks was 260.
jan 19 aiftp: 232
feb 19: 238, predicting 250 3/19.
3/19: 260 (which is where i was last fall)
new prediction for 4/16: 261, which seems like a vastly different projection from the previous 2 increases.
Restarted TR a month ago, AI had me at 277 which wasn’t remotely close. Moved me down to 260 watts today which is dead on. Very happy to have it updated and accurate now
I have no idea where 277 was from, I did have a 20 minute effort close to that a couple of months ago but that was all out
Feb 19 224W prediction
adjusted calendar for 5 days of travel, new predicition 219W
Completed all the workouts on schedule
March 19 FTP Detection 219W.
Funny, it was predicting 220W the day before and after my final workout. But when the detection went live the following day (a rest day) it went back to 219. No real complaints EXCEPT that maybe the system is a little too precise when I’m seeing 1-2W dips during the course of a week when I’m hitting my targets and feeling good. I know those differences are meaningless, but I can see how it would be tempting to rate workouts easier to get those couple of watts, which of course, defeats the purpose of the survey.
This 100%. the number doesn’t matter, but the 2-3 W adjustment literally days before the AI FTP detection just seems to sensitive. My 4 week prediction is usually about a 3-4% jump and ends up being 2% for a variety of life-isn’t-perfect reasons.
Example, followed the plan to a T for the past 4 week. Had a planned FTP of 289. Woke up at 2:30 am on Thursday but eventually got up and got on the bike at 4am fo9r Mount Alyeska. First suprathreshold interval was tough and backpedaled through the last 40 seconds on the 4 minute interval. Nailed the subsequent five intervals no problem but it was tough. Rated it hard and predicted FTP for today went down 2W, due to the backpedal? Did a group workout in a training center last night, more sweetspot and that pulled the predicted FTP down another watt this morning.
Been a loyal TR user for 14+ years and encouraged so many friends to use it because I believed in it and my engineering mindset just bought into it, but this is the most I’ve questioned how well this is executed. I have so many of those friends now questioning the wildly high or even decreasing FTP predictions they are seeing. They are losing trust in the product.
Sometimes you need to adjust the commercial messaging of a product or feature to get buy in from your customers. It seems like the commercial adoption has missed the mark here and they may actually be revealing too much to the user to get the buy in on the AI predictions.
Yeah @agreif , thinking about a practical solution to the sensitivity, maybe 3-4 weeks out the prediction gives you a range , ie FTP 220-224 that assumes 100% compliance and survey ratings matched to predictions. If you fail or miss a workout, maybe you get an alert that says you’re at risk of falling out of that range. Then maybe the final week you get hard numbers.
TLDNR: The Ai works, but it’s giving our dopamine-addled human brains too much information 
Well 2nd block on the new AI is now complete.
My prediction from Feb 18 was 324W
I ended up getting sick weekend before last, and skipped my Monday V02 workout of that week. Wasn’t feeling great later in the week but managed to push through and get my other workouts done.
New AI FTP detection this Wednesday March 20: 320W
The first sweetspot workout that same day with the new FTP was not easy, my heartrate was a bit higher than normal but I also haven’t been sleeping great and considering I’m pushing watt levels that would have been appropriate for a V02 workout 6 months ago I’m pretty much thrilled.
My new prediction for April 15: 330W
Had a hard-start threshold workout tonight, and I also added 15 minutes to my normal workout length, it was pretty brutal but I did get it done. Definitely “Very Hard” even though it was predicting hard. Although…maybe I could have got one more interval done? If I really had to 
Didn’t help that my normal Thursday night hockey game was short players last night and ended up with some extra shifts.
Anyways, this has dropped my prediction down to 328W, which I am totally ok with. With the nicer weather (hopefully) coming soon I’m excited to up my volume and see what I can do outside 
AIFTP update - 246 March 22, 2026. ( Went up even with rating two workouts Very Hard 
)
FTP prediction - 249 for April 19, 2026
It detected at 261w in the end which is probably a bit high and for the next prediction its giving me 272 w (4.37w/kg), we’ll see in the next few weeks
Some time in Jan I got 243. I declined and went with my Nov detection of 234 because I felt I hadn’t don’t enough training to warrant any higher.
25 Feb - AI FTP accepted at 244 and predicted somewhere in the 250s, I think 253-255 or so.
25 Mar - AI FTP detected at 251 and predicted to 268
Getting so much stronger!
The prediction honed in on about 251-3 over the past week or so. I am expecting the 268 to decline a bit but will be extremely happy if I get that.
If I lose a couple of kgs I’ll be getting very close to the 4wkg club. Progress is not linear but it looks achievable for the summer!
3-6h a week, of which 3 hard workouts and the rest varies if I do group rides or whatever. In early 30s with v stressful career (long hours, travel, high stakes). Very happy with progress since I leaned in to the new AI model.
I don’t want to do workouts with an FTP of 349w 
Usually the optimistic predicted FTP has dropped to something more sensible by now but I’ve only got 5 workouts left to get the FTP down to an acceptable level…. wish me luck 
Ha! Having the same problem. Handed every single aspect of my training completely over to TRAI in Erg Mode and seeing some great improvement. Problem is now I have to train at these new numbers !
Climbing on the bike for Cloudripper in about an hour.
Pray for me
“NEED HELP: FTP prediction keeps going up and I don’t know what i’m doing wrong” 
