Prescribed strength training would be awesome but far too much variability in tracking that data and relying on an RPE entry to truly progressive overload someone. They would need to gather so much data on strength training and it would be incredibly messy for the model.
I’d be happy with slightly more predictive plans. Namely, I tend to pick a race, put a plan on, follow for a while, then something happens and I deviate. Would be nice if the platform just told me “you always skip your Wednesday ride, so we’re not even going to program it” - you blow up past 15 hours a week, we won’t prescribe it, you always struggle to hit threshold reps on days where you ran the day prior. Just realistic coach like feedback, but not a generic LLM output that has no backing.
I’m sure whatever release is coming will be their best effort of adding an LLM element for customers. Perhaps workout descriptions that have a positive correlation with athletes breaking through hard days? No clue.
More to the point, they have no competitive advantage here. No real data, not a ton of experience. It just wouldn’t make business sense no matter how much some of us would love it. At best I could see improved training plans that take concurrent sports into account in some improved way.
TR’s competitive advantage is their platform, and they could add support for strength training and logging. Training plans could be created by strength coaches they hire or partner with. Even if they added fixed strength training plans and/or yoga sessions, that’d be something. Ditto for nutrition.
Imagine the system was able to analyse your metric within workouts on both zwift and the app to change interval difficulty on the fly to give you optimal metabolic system targeting. Ie you’re doing this SS workout today and your heart rate is much higher and cadence is lower than previously, you’ve done a lot this week, lowering the intensity would result in more meaningful adaptations at a more achievable level, preventing burnout. Additionally increasing session intensity based on your metrics (from previous training), ie a lower than normal HR with high reactivity to changing intensity, relatively low CTL coming into session, Good cadence, etc.
intensity changes would also be based on pre workout surveys. And entirely optional, like “hey you seem to be crushing this workout, shall we add Intensity of the working sets?”
I like the idea of this, and i hate it in reality. When i’m suffering i get this absolute barbaric ‘i have to beat this’ mentality and it usually gets me through a set. If the system just lowered by output because i was having a shit day i’d feel a bit cheated.
I guess if the system is designed to just make me faster and it proves it will do so, i’d be behind it.
Another idea of feature that is within the realm of possibility would be ‘route specific plans.’ TR has enough data to analyze a gpx or fit file from somebody elses ride, understand the demands of that ride and build a plan that would exploit those demands for the best outcome. Crit racer with 700w surges out of each corner would get very different training than enduro gravel bro who just has to hold 300w for the flats and surge on climbs.
I’d absolutely love a feature that let me tag segments from strava and have it prescribe training and weight loss suggestions to help me improve at that specific segment. I’ve anecdotally done this by attempting to improve my 2 min power and 20 min power for respective segments i want to beat but it would be neat to have sessions that incorporate the segment itself. Text on screen could be good cues to help a rider mentally prepare. ‘this interval is going to help when you hit that 18% part of the xyz segment that you’re training for - you’ll be hammered at this point, but its what you need to improve at if you want to beat ABC’s time’
TLDR: gamification tied into strava segments. I eat that shit up
Yes me too. It wouldnt be automatic but an option during rest periods between intervals or with an easy override option. Kind of like the directional selection method on zwift.
I get you, but again feel is more subjective, my hypothetical scenario would be completely optional, it also begs the questwhy you need TR at all if you have a brain. Just employ it for all aspects of your training plan
The CP model’s range of validity is too small, because it is too simple. A ML-based model will have too many parameters and none of them will have a clear interpretation.
You’re spot on, the next couple of months I bet sees a peak of the amount of people training compared to last couple of months and a whizzy new feature would attract new people.
It won’t be these but the triathlon plans feel like bike plans without much thought given to the running and swimming. And even if they don’t prescribe resistance training they need an option to make some of your weekly workouts resistance training and account for that in the weekly hours and fatigue.
They must have a lot of difficulty making plans or recommending training for anything that isn’t a bike workout where they have the power data.
