// TOPICS COVERED
(00:00:00) Welcome & Why TrainerRoad AI Is a Major Update
(00:02:00) How TrainerRoad AI Has Evolved Over Time
(00:04:00) Why TrainerRoad AI Isn’t Just a Chatbot or LLM
(00:06:30) How TrainerRoad AI Simulates and Selects Workouts
(00:10:30) How TrainerRoad AI Replaces Static Training Plans
(00:15:00) How TrainerRoad AI Reduces Failed Workouts and Burnout
(00:20:40) How TrainerRoad AI Adjusts for Fatigue and Big Rides
(00:32:00) Why Most AI Training Tools Don’t Validate Workouts
(00:36:10) TrainerRoad AI Training Forecasts and Simulations
(00:49:40) TrainerRoad AI Workout Alternatives Explained
(01:03:20) Why Long Rides Can Undermine Progress
(01:17:20) Conservative vs Aggressive Training in TrainerRoad AI
(01:37:30) How TrainerRoad AI Changes How Athletes Get Faster
In this episode, Nate and Coach Jonathan explain all the details behind TrainerRoad AI, the biggest evolution yet in TrainerRoad’s training system, walking through how the new AI-driven approach goes far beyond static plans to dynamically simulate, predict, and personalize every workout on your calendar. They explain how years of performance data, workout feedback, power and heart rate, and progression history now power a system that actively chooses the right workout for the day, reduces burnout, cuts down workouts that are too hard or too easy, and helps athletes recover faster from missed sessions or failures. The conversation dives into how simulations work behind the scenes, why long rides and “hidden fatigue” can quietly sabotage progress, and how features like AI Predicted Difficulty, AI Training Simulation, Dynamic Duration, and Training Approach sliders give athletes confidence that every session is worth their time. The result is training that feels consistently “just right,” adapts to real life, and helps athletes get faster with less wasted effort and fewer mistakes along the way.
I had set up a consistent training plan to raise ftp a week or 2 ago and now Ai says if I do all these workouts my ftp will actually be lower than when I started. (By 1 watt) but I also gained a free 10 watts with the new Ai, which you mentioned might happen in the last podcast.
Also. Seems like the next step is to incorporate generalities based on people’s bloodwork. Looking back on years of riding I can see that I feel proper fresh like 2-3 times a season usually after forced breaks for work. I don’t think this is normal. But I NEVER feel fresh. Even after vacation. Rest and food. Nothing. Feel flat. I’ll be strong but NEVER fully recovered. It feels like deep fatigue sometimes. I’ll do everything right and my legs just won’t come around then out of nowhere they do. Often the top end just not there but I can keep on trucking. Anyway, I finally get bloodwork done and it’s showing high levels of SHBG, which is gobbling up all my free testosterone, which has a massive impact on recovery, muscle growth, fatigue, the list goes on. So, is the TR training plan correct for my ftp and the average bell curve athlete with that ftp? Yes. Is it correct for someone with that FTP and high SHBG? No. You see where I’m going with this?
Maybe Ai will be able to sniff out my actual recovery status based on my performances even if my homeostasis is to be in a seriously fatigued state and figure out a way to boost ftp and increase recovery with bad blood or will it just drop my ftp. I suppose my ftp would just be a lot higher if I had good blood. It seems hard to get past ftp as the cornerstone of a plan because it seems like the model suggests if your ftp is X than you should be able to do Y. My ftp is X but I can’t do Y all the time. The problem with most ftp tests is that they have nothing to do with peak power, which fatigue seems to hit hardest, and you can be seriously fatigued and perform quite well in an ftp test.
Seems like the bottom line is, trust the system, do the workouts, forget what it says your ftp is
I could keep going but. My kid is almost 2ish, I’m 46ish, 300ish ftp (320ftp with new Ai, I’ll take it) 160ish lbs. Crazy how old that sounds. I can hear 50 making fun of me. FTP has been the same for years despite my efforts. I use TR pretty much winters only. I ain’t going to the Tour de France so don’t care if I’m leaving watts on the table by not picking up the hottest trends and changing my crank lengths and cadence and whatever else. But fairly structured riding and usually test out a new approach every season. But when the weathers nice I’m gonna go as far and as hard as I want. Don’t care if it’s not the best approach for whatever.
Coming off of a layoff and doing volume, FTP detection had pegged me at like 314, then after I was added to the AI group, it dropped to 294
This was definitely more in line with both my perceived effort as well as expectations.
Just last night at the end of a TdZ event, I had a 21 odd minute climb JUST over 300 and I was pretty cooked.
I’d say this is looking pretty accurate.
Love seeing my progression levels based off of full history too, despite not having done TR workouts specifically that would have affected them.
I’ve said it in the beta thread multiple times and have sadly been ignored by TR staff: I’m open to training under the new system which raised my FTP from 308 W to 342 W, but the whole system now uses the wrong training zone names for all people with such significant FTP changes.
Over-unders are actually over-overs, sweet spot is actually threshold work and threshold is VO2max in reality. This irks me to no end and it introduces errors in other systems. E.g. my Garmin takes the TR FTP value of a workout as truth and calculates a wrong TSS from it!
Garmin (wrong FTP 342 W inherited from TR): Load 69 TSS
Nobody at TR had responded to these specific concerns ever and I’m tired of being ignored here.
The traditional FTP has a useful meaning for me, and all other platforms agree to within a few percent that it’s around 300 W for me currently. I don’t mind training with 342 W instead, but this has so many major repercussions on how workouts have to be named and processed into other systems that still have the traditional FTP meaning.
It seems to me that there is a huge error in the AI FTP calculation. I did a TT race last Sunday, and the result was 256W for 45 minutes. That was the maximum I’m capable of. I couldn’t do it any harder. So I set my FTP at 256W. Although today’s AI release has made it rise up to 271W! Where did it get it from? The most important thing is what am I doing about that? Should I set the realistic FTP, giving up a part of AI functionality? Should I keep it that high, failing workouts?
I opened the app this morning to see that it has been updated. A layman’s question - my FTP has been reduced from 270 to 252 (a 6.7% reduction). Forget the ego hit for a moment (not easy!), but given that I was nailing my workouts based on 270, does this not mean that I’ll find workouts going forward too easy? Also, if I were to complete a ramp test tomorrow and my FTP comes out at or around the 270 it was previously, will my workout difficulty be upwardly adjusted accordingly and will the FTP on my profile revert to this number?
I thought I would comment here rather than starting a new topic.
Okay so, up until tuesday 20th of january, my FTP was 266W according to TR AI.
After the workout on tuesday, FTP detection bumps it up to 277W. Okay.
Now that the new features are being rolled out, TR detection increased it again to 285W.
I was finding the workouts to be reasonnably challenging, so my question is, am I going to explode on that trainer ? Is it a tiny bug ?
Also in the top right corner it still says 266W.
Most of the people in the beta who had their FTP drop found their workouts got harder. Sounds counter intuitive I know. Basically you would do lower power for a longer time. Instead of doing 10 minute sweet spot intervals they were doing longer intervals at a slightly lower power or similar power 20 @ 94% vs 10 @ 90%.
The folks with the FTP increases had their long intervals decrease, I was doing 20 to 24 minute sweet spot at 94% and they dropped to 10-12 at a slightly higher power but lower percentage of my FTP 86-90%.
Give it a try. I think you will get good workouts and results. If the system is off it will adjust as you go.
I have many doubts on the new FTP detection , I m resuming cycling training after christmas break (it was almost 2 months out of the bike, just running and lifting), and last week the FTP detection gave me a 234 W FTP. that i think it was realistic. yesterday after the update, the new FTP detection upadate my FTP for 278 W. There s no way in this world that I can sustain 278w. there any possibility of run the ftp detection again?
My FTP dropped to the lowest it’s been in a long time after the AI integration. I was at a point in my training plan where I’d started V02 max and anaerobic workouts with varying degrees of success (V02 good, anaerobic not so much). To me, FTP was just a number, but you guys have preached how it’s the be-all end-all for measuring training success and it’s disheartening to know the work I’ve put in to this point and to see that number drop markedly. I’ve decided to bail on that plan since it’s obviously not making me faster.
You’re bailing when literally none of the workouts you’re doing should really be prescribed based on an FTP number, whether you agree with the number or not.
I feel that when I schedule my week using TrainerRoad AI — for example VO₂max on Tuesday, endurance on Wednesday, tempo on Thursday, endurance on Friday, over/unders on Saturday, and dynamic endurance on Sunday — the endurance sessions keep progressing in level.
These end up being around 0.75 IF for 1.5 hours, which is actually quite taxing. The intention of those mid-week endurance days, at least from my perspective, is to stay fresh and ready to perform in the key workouts.
My point is that the AI should adapt by limiting progression in endurance workouts in this kind of setup, because pushing endurance too much will, in my opinion, reduce freshness and negatively affect my ability to complete the higher-intensity sessions effectively.
Same here. I have no interest in doing endurance rides at 74%. I set my endurance training zone to conservative instead of balanced and it gave me the same length workouts but at a much more manageable percentage…for me.
