Has anyone else noticed this? I’m a long term user of TR from back in the day so have seen it change a lot. Something I don’t like about the new AI version is that when it changes the next lot of workouts in the background after a workout has finished it doesn’t tell you what it’s actually done and how it’s changed your future workouts. The previous TR version at least could suggest changes so you could see what it was trying to do whereas the new AI version you are really training blind in my opinion. To me the whole point of TR and a lot of its value has been that you are informed about your training as it goes on week after week. To me the new AI version fails in this area. I’d love it to give me more info after changes about why it’s doing it. anyone think this?
Unless you look in the calendar?
That’s a wild opinion imo - I know what my training is like constantly.
Your week is structured, it fluctuates slightly in detail according to your performance and feedback. I
yes it is constantly changing stuff without flagging what it has done now, which makes it impossible to get a sense of whether it has adjusted you “up” or “down” as you go. but then seeing as ftp also is now just some shifting variable that no one really understands (some floating functional number they use to calc workouts) it’s all a bit mute anyway
Pin your workouts as they get populated in the simulation window. Then on the day, unpin them to see what the model is now suggesting as a result of what you’ve been doing.
I get that but you can’t see the changes to the ongoing plan now like you could before. To me that’s a negative in the context of understanding the training that’s taking place.
I understand what you’re saying, but it literally recalculates the next 28 days every single time you make a change. That amount of ‘why” data would be overwhelming. The way @Flashpoint51 pointed out would work for “day of” changes though.
Yep. That’s exactly my point. I don’t think the new Ai version is actually helping people understand what the long term plan is doing. I think that’s a backwards step in some ways.
You could do that but it’s a bit of an afterthought. To me the very definition of ai is the understanding element. I think TR has missed a trick here.
Yesterday I inadvertently moved A workout. Now the guessing game was what did I do to get it back.
on one hand I see the desire yet if you a lot of moving and playing around it is pointless just due to the volume of changes if you do this multiple times.
Changing ftp manually and then doing a ftp detection is all recorded on the calendar. It is a pointless recording if you actually never do a ride at the different ftp and just want to see the changes.
The only changes I have experienced is very slight structure changes to workout. When I look ahead, I know that I am going to get over/under threshold changes on Saturdays, and maybe by the time I get there it picks a very minor increase/decrease to the workout depending on how my training has been going, but I don’t need to see every little change. I trust that it is responding to my actual performance and feedback and giving me appropriate workouts based on that.
Personally I think you get a broad understanding of the long term plan by looking at the structure of each week through the phases. As @KonaSS says, changes to hard sessions are only really going to be intensity or interval/recovery duration related - unless you run into fatigue detection, when it’s very evident what’s going on.
In any case, Nate has talked about using an LLM type feature in the future to add context, so I don’t view it as a missed trick, as much as the latest iteration of a product that’s undergoing continual development. Don’t let the perfect be the enemy of the good.
If you move something and then put it back it allegedly should come up with the same calculations it did before and put everything back how it was. Someone from staff said this but don’t have a direct link to who or when it was.
Time is a factor in the system though both test and even potential time of day you might do a workout as noted by Jonathan on the last episode with hannah. So if you change it in the morning and put it back at night then you might not get the same result as the system may have already recalculated.
Yep. I don’t like it at all. I know we trust the procedure and all that but I think if you lose all ability to monitor and understand what’s happening then that’s a step backwards IMO
Personally I’d have preferred they leave it as it was until the iteration was further down the road.
Yep but before you could do a workout and you could reject changes based on how you perceived you felt. IMO that was a good balance of ai and user input.
Still disagree with your summation, you can monitor and there is no more or less understanding than before.
Perhaps you thought this was an LLM? Check out recent podcasts to hear it from the horses mouth.
Actually there is more understanding from the predictions and power curves.
Are you really losing all ability to monitor and understand what’s going on? Seems pretty simple to me. The prescribed workout duration and type (e.g. 90 minute over unders) is pretty fixed according to the plan you set up and only changes if you add in enough unexpected fatigue that it feels you need a recovery day instead of a hard workout. If I’m completing workouts broadly as prescribed and my RPE is as predicted then things just tick along with maybe some incremental changes. If I’m going better than predicted (e.g. I’m significantly above power targets on a hard interval session but RPE and HR are as expected) then it may dial things up a bit. If I’m skipping or failing key workouts then it dials me back. If I chuck a load of extra volume and fatigue in there that is going to impact a key workout then it also dials it back. Pretty much what I’d expect. Not seen any major changes that I can’t explain.
If anything what I’ve seen has led me to have more trust that the AI knows what it’s doing and take a bit more effort to therefore tell it what I’m planning to do. E.g. If I’m going to ignore the recommended 2 hour weekend ride and go do a 4 hour group ride then I’ll actually stick that on the calendar in advance, see what it thinks the impact of that will be and then adjust accordingly. Which might mean moving an interval session to later in the week to avoid a yellow.
There’s definately not as much instant info on what has changed in your workouts ahead than there was in older pre ai version. That’s a fact as it just doesn’t tell you or show you that now. Yes you could drill down into other elements but that’s not always practical as time is a factor sometimes.
I equate to a bit like going for a ride in a group but you’re the one who hasn’t got the route on your head unit. Yea you’ll still get there but it’s better if you’ve got the route.
At the end of the day some like it and some don’t but it is a fact that there isn’t as yes post workout at your fingertips info as there was before. Hopefully that’ll change going forward. I think it will as the first pods from Nate after the change seemed to indicate the ai would be a lot more personable in the future.
Well, it doesn’t pop-up “accept changes”, but it closes a workout into the calendar where you see any changes being generated. You know what workout type/length, and you know when it’s changing. It’s just different to the user journey before.
Again, I can’t possibly agree with this summation - you absolutely do know what workout you’re doing before you do it. So you’re turning up for the ride with the route on your head unit.
If I asked my club what route the Sunday club ride was going to be in two weeks time, they’d laugh. 