TrainerRoad’s Biggest Update Ever Is Coming 👀

[Nate_Pearson]

@JMGant

Beyond stoked for this! 600+ hours of TR training in 2025, and I have seen great improvements in my performance. For my training, I lean heavily into my prescribed TR workouts, and I am always looking for that extra power increase. I am looking forward to seeing the updates and improvements to an already stellar product.

Thank you!!! I’m SO excited about this one!

[Nate_Pearson]

@craigmanning

27% less likely to adapt your workout to Pioneer-3?

Yes!

There’s a new feature called Workout Extension.

What was happening is Pioneer-3 is about the max sweet spot you can do for an hour. Now, if you’re at the top of your progression level for that duration, it will increase the duration of your workout with the ability to swap back.

[mjw0]

I’d be happy to have Apple/Android Health just to be able to sync workouts to a calorie counter app without having to loop in a third party like Strava.

[Nate_Pearson]

@urnowsmoking

So it can look at your plan and the current and future workouts and predict how u will end up regarding ftp and other metrics? So basically what I requested and posted over a year ago as to what was possible with Ai but the community roasted me.
I can dig up the posts and the nay sayers names if anyone likes!

Yes!

[Nate_Pearson]

@Dexter

What’s the purpose of announcing ‘biggest ever AI improvements’ and then not event saying the date it will be rolled out to existing TR users?

It reminds me of those announcements of GPT 5.2 or some expensive movies where the expectations became too high after the announcements and the reality was not meeting the expectations (hope I’m wrong here for TR)

Because the biggest training day of the year, Jan 1st is coming.

But, if we announce Dec 31st, and the team has a critical bug and we miss that, people will be SO MAD and confused.

We’re really in a damned if we do and damned if we don’t.

I’m confident that it’s our biggest AI improvement by far and our metrics look great and our beta feedback is great.

[Nate_Pearson]

@kevistraining

Feels like they had their heart set on a Christmas day announcement

We set it so we had content for each day leading up to our launch. We didn’t say the specific date because of what I said above.

[Nate_Pearson]

@kevistraining

I don’t think there is anything nefarious going on…

I’m sure it’s pretty much done and they just want to run it past a few more beta testers first.

There are just some critical bugs that the beta testers found, so we’re fixing those.

When they are fixed we’ll probably launch the next morning. And we might include more beta testers the night before as a final check.

We’re not going to do a full week of adding beta testers and let them play with it for a week for feedback.

[Nate_Pearson]

@NelsonBretts

Cant wait big big fan of the current system so anything more is a BONUS!!!

Question, if I go for a random gravel ride or a spin on my eMTB, will this new system still update my training plan, even if I haven’t scheduled these outdoor sessions?

100%!!!

[Rizzi]

@Nate_Pearson

It’s not HR based in terms of guiding your power with HR. But it does look at your HR and use it in your simulations and thus your recommendations.

BTW: Could you imagine to incorporate something of the new system into the workout player in order to get real time feedback / adjustments (xert workout player comes to mind or the xert Garmin app)? Or is that too far fetched?

[lp251]

Any plans to hire more MLE/ml researchers? asking for no reason in particular

[Nate_Pearson]

@NelsonBretts

So if I have a planned week and then do a outdoor unstructured ride that I can’t link the adaptations don’t count (I think ) also only if I mark it hard will it change a workout planned the next day or if I plan a outdoor unstructured ride for a Friday say and then have a sweet spot session Saturday it will stay even if I estimate a tss for Friday.

maybe I’m over thinking but I don’t think I would be in best situation on Saturday and would probably impact real progress? Maybe.

All of those are joyfully wrong :-D.

1. If you don’t link an unstructured outdoor ride it will still progress you in your progression levels based on that ride’s power. Or it might step you back in the next workout if you need more rest between workouts. It doesn’t mindlessly step you forward.
2. If you plan an outdoor ride on Saturday, the planned Sunday sweet spot ride will change in response to your planned Saturday outside ride.
3. You’ll be able to see how Saturday’s ride impacts your fitness and predicted FTP BEFORE you do it!

[Nate_Pearson]

@ChefAcB

I personally got burned out from TR mentally by seeing over -unders every base, build and specialty.

We don’t have them every build, base, and specialty. But we do use them a lot because they make you faster.

If you don’t like them, I suggest changing the workout profile on those threshold workouts to another threshold workout profile. You’d just have to change 4 workouts a month at most; less than 45 seconds. This is supported in our new launch.

[Rizzi]

@Nate_Pearson

If you don’t link an unstructured outdoor ride it will still progress you in your progression levels based on that ride’s power.

So this means WLv2 (workout levels for unstructured outdoor rides) is real?

So do unstructured outdoor rides get a progression level for each zone?

[Nate_Pearson]

@WillLocke

Despite all the new features, I don’t see any claims that the new AI training platform is more effective than the adaptive training approach. There are reports that the new platform is more likely to select the “right” workout, but what is “right” is measured by the athlete’s subjective opinion. It’s a big leap to say that doing workouts that are more often subjectively good will change performance outcomes. Maybe that’s coming after the beta finishes?

The teasers also don’t mention any support or integration for weight training, heat training, running, sleep/HRV/RHR (why can’t we connect our wearables to feed this information to the platform?), lactate testing, or custom zones (still stuck with FTP-derived 7 zone model?) that are pretty much standard in modern performance cycling. I really hope they develop a product that supports these aspects of training.

I’ve responded to most of these already but I’ll hit the highlights.

1. We were going to post some FTP increase data but the people who needed to do that had to finish the model/data warehouse work then went on vacation for Christmas. I suspect if we did post FTP increase someone would respond with how FTP doesn’t really matter and we concentrate on it too much (as I’ve already seen in the forum). When we launch, you’ll be able to see your OWN FTP increase based on what’s in your simulation window which is pretty cool!

2. The right workout is not just subjective opinion. Two examples are not being able to complete a workout’s target wattage, “super passing” a workout where you do way more than what was prescribed, or having a really low heart rate on a VO2 max workout where you’re supposed to get your HR up.

3. Running is integrated into the AI model and improves it but we can get more sophisticated with it for future updates. I did a post above about wearables.

4. The AI just knows watts and HR, it doesn’t understand FTP, lactate testing, zones, or any of that. It just understands what you’re actually doing, and then can tell you what you need to do to increase you watts. That’s basically the root of cycling training, and the other stuff around it is just used to help people understand it. But with an AI, it’s developed it’s own complex statical analysis that’s more complex than a human can understand. I go into this in more detail on the next podcast.

After the AI picks the watts you should do, we translate that back to FTP and PL to give athletes a mental model of what they are doing.

[Nate_Pearson]

@BAThomas69

A follow up question. Will we get an email notice if we’re accepted in? I’d assume so, but don’t want to be in a “you know what they say when you assume” type situation.

I’d also guess when @Brany said “very soon” and the fact that @Jonathan said Nate will give more details in the next podcast, that probably means a light rollout starting early next year. And depending on how that goes, then more and more waves, each likely getting slightly larger. I remember hearing them say when they were rolling out Adaptive Training, they were doing it in waves.

I’d think it was random, though they may also have a certain percentage (probably a large percent) each wave of people that meet the criteria that they are looking for.

The plan right now is to make it early access available for everyone at once and give it to new sign ups.

We wanted to give it to everyone at once but we have some people out for the new year and we want a fall back incase there’s a critical bug in an area where we have light coverage.

[Nate_Pearson]

@craigmanning

So now every time you open TR you will be consuming a water and natural resources.

We’re not an LLM, so what we’re doing is a fraction of the amount of resources compared to that. Way less than you using google or watching youtube/tiktok.

[LacticacidMCB]

@Nate_Pearson

If you don’t have power, I’m pretty sure it still reads in HR only but would be less accurate, but I’d have to double check that.

So now I have an excuse to get power on my mountain bike.

[Nate_Pearson]

@hubcyclist

my feeling is this is even more overcomplicating of something that is relatively simple. for me training can easily be broken down into a few things a) building endurance and volume b) building TTE c) increasing aerobic capacity and d) sharpening anaerobic capacity (and I guess sprinting, which I don’t care about lol)

while I know athletes are different, at the same time I don’t think we’re all unique snowflakes needing hyperspecific training and constant adjustment on the fly.

If that were true, we’d have great results with a static training plan.

And I think we’ve all experienced a VO2 max workout where 2 min rest between intervals was good, but 1:30 would have buried us. Or 5 more watts on a long threshold interval would be too much.

The nice part about this release is it actually simplifies things. You can just open the app and train and close it, everything else is handled for you.

The other features aren’t required to use the product.

[Nate_Pearson]

@JustinLavish

I don’t trust my Garmin watch’s sleep and HRV data enough to put it into TrainerRoad’s algorithms

We plan on bringing these in and then validating it. The AI will tell us if it gets us better predictions or not.

[hubcyclist]

you guys need to get a lot less prescriptive with the vo2 stuff, recommend targets based on the user 3-5min power curve, allow for flexible rest between intervals, etc. I will easily do 2-3x interval duration for rest between vo2 efforts