TR FTP input has problematic for a ton of users far longer than FTP detection has been around…a quick search of debates on the different FTP assessment methods will show how deep that rabbit hole goes.
The workouts and plans as initially built were based off a classical FTP type approach. This assumes 1) that TR users know how to set their FTP; and 2) that they know how to adjust intensity from the prescribed targets.
The ramp test was introduced to help with 1, and kinda did, but from my viewpoint drifted farther away from getting FTP accurate. The game changer was developing progression levels where knowing FTP was no longer super important - you can have a much wider margin of error. This latest development on FTP detection is a natural next step, since really what matters is how you perform on workouts and if you are progressing more than anything else.
Instead of getting Goddard -4 as a default workout replacement when you opt for AI FTP Detection instead your Ramp Test, athletes will now get a replacement workout custom to their plan!
Thanks for continuing to provide feedback to the team.
I’ve argued a lot that AT can handle a wide margin on the FTP input, so I’m gonna put some skin in the game and accept the new number.
I am confident that I can work with the AT recommendations to yield a highly productive training stimulus, despite the fact that the number I got today is way too high for my 40-70 min power capacity, and I think my current input - that I got 14 days ago - was a better number to use.
For what it’s worth (and I shared this elsewhere in the thread), AI FTP Detection nailed my FTP after several months away from the bike. Coming back to training postpartum, the last thing I wanted to do was take a Ramp Test. I guessed my FTP at first as a thought exercise, figuring about a 60 watt drop, and it was too high (I immediately failed the first workout I attempted). AI FTP suggested 10 watts lower than my guess, which ended up feeling perfect. A full 70 watt drop in my FTP.
As has been mentioned, folks are more likely to share their increases than their decreases here, but in our internal testing, AI FTP detected detraining as well as it detected gains.
We just fixed this! Now, if you have more than one workout scheduled on the same day as a Ramp Test, we’ll automatically prioritize the Ramp Test on your Career Page, to give you access to the AI FTP Detection feature!
Its EXACTLY 4% for me too WHICH IS HUGE. Thats a 10W difference
Its easy to think everyone is probably running with over-inflated actual FTP - I wonder if that’s the case. It would be a cool to see the avg PL for TR users, especially in threshold. Is everyone doing PL lvl1 and 2 threshold workouts with AI FTP and working up to 6+?
If I was to accept AI FTP it would mean my Over’s in O/U would be at v02max and completely unsustainble.
I always prefer to have a lower FTP anyway - helps me look forward to workouts as I know they are achievable and if I’m feeling good I can always exceed the recommended wattage.
That is exactly what I have found… and my point above is AT can’t fix that… it just pushes shorter and shorter intervals at you which you also fail until the intervals are so short they are totally unproductive. Instead of getting workouts with 30 plus minutes TiZ around your Threshold you end up with a workouts that are labelled as Threshold but are supra-threshold and short in duration.
I’d say you need a realistic value within about 2%… I’ll see what it throws out next time but if it repeats the same margin of error its not usable for me.
WHICH IS HUGE. Thats a 10W difference