First AIFTP change

Been working my short custom plan* for 4 weeks now, so got my first “do you want to discover your new AIFTP result” prompt.

I won’t lie, I was in two minds. It’s Friday and I’m in a good mood, seeing a drop would have really pissed on my bonfire :grin:

Clicked it, and boom, yes, +5W, 271 → 276. (Weight of 73kg).

Happy with that.

I AM curious how it works though, what factors dictate the number? I’m sure no-one outside of TR truly knows, but what’s the community consensus?

Do we think there’s a clever algorithm that’s comparing our completed workouts with those who have done the same workout and a recent FTP test? Assessing things like HR and HR recovery rate?

Or is it in fact JUST the post workout surveys?

Or does, in fact, everyone get a 5W bump the first time around, just to give us a warm fuzzy feeling and the belief that TR is working. That first hit that we’ll always want to chase?


*only 3 months to my A race, so it gave me a couple of weeks of base, a recovery week, and now I’m in my first week of build. Which I’ve rightly or wrongly (wrongly, probably) added an hour of endurance to every day, since I’m going on holiday next week…

1 Like

One ITEM we have from TR is the loose info in their support site:

How does AI FTP Detection work?

We used Machine Learning (a subset of Artificial Intelligence) to analyze millions of rides, to understand how FTP changes over time and in response to training. Our system looks at your personal biometrics and the unique details of your recent and past cycling data to detect your FTP without the need for maximal efforts or specific workouts.


Beyond that, you can check the comments from Nate and a few other TR reps in the intro topic that shed some more light:

  • Unlikely to be just this, but I believe is one of the factors.
  • Highly unlikely since that doesn’t parallel my own experience, not that of others who got DROPS instead of increases.

My experience with AI FTP:

  • Nov 22 - Ramp test 248W. This comes on the tail end of 6-7 weeks of very little riding, probably down from a ~270-275W FTP.
  • Dec 8 - AI detection confirms 248W
  • Jan 4th - AI detection raised to 254W

I started an adaptive plan Nov 23rd, which had me do 3+1 weeks of General Base 2 at 7-8.5hrs/week, long rides done outside without power. Following that I had 2+1 weeks of General Base 3 at 10 and 11.5hrs, long rides done outside without power again. 254W AI detection was done in middle of the recovery week.

Is it accurate? I have no clue, I may do a ramp test on Monday to compare, will update if I do. I’ve certainly improved my endurance/tempo, but being in base, I have stayed in threshold or below, so I’m unsure how much change there would be with my FTP.

edit: My long outdoor rides (1-2/week) do not have power data. I’m not sure how much that affects AI FTP detection, if at all. I am assuming effect would be minimal since I still have 4-5 weekly rides with power in various zones (endurance + threshold).

(I should have said, pre starting this TR block I did 3 zwift ramp tests 3 weeks apart - 262W, 267W, 264W, so the initial AIFTP figure of 271W felt “about right”.)

1 Like