🔪 AI is killing my rides

But you did have to note, twice within a brief comment, the importance of a “properly ascertained” or “solidly set” FTP. Too many people don’t do long-interval testing, and some with a strong anaerobic response get their FTP rated too highly on TrainerRoad. So 60 TIZ is definitely not a given in this population.

Nothing wrong with a PL 1.4 workout for sweet spot. Some people ARE just getting started, their FTP is NOT clearly and solidly known, or they’re recovering from injury… whatever.

The issue for this thread was not about the PL 1.4 workout being good or bad, but about someone who had scheduled a PL 4.0 workout and AI suggested an alternative that was 1.4, so they asked why such a change would happen and how to respond.


getting the right FTP isn’t that hard though, even if one were to be a complete novice, I tell them try the ramp test, we knock down a few percentage points to account for overestimation and have them attempt a 30+min effort, it’s like one extra step and would serve people way better. so, with all due respect, posts like yours kinda come off as hand waving to justify kinda mediocre training principles

1 Like

Uh-huh. And just how is someone without a coach, or who doesn’t frequent the forum – and MOST of TrainerRoad’s clientele is going to fit both those criteria – going to know that they might be one of the people whose FTP is overestimated, and to try knocking it down?

How are they going to know that what feels really hard about Sweet Spot to them… shouldn’t?

1 Like

The internet is an amazing place with lots of info. And TR could easily bake a different onboarding experience into their programming if they really wanted. Again, hand waving

1 Like

That’s on tr? Have they not changed and evolved?

1 Like

Sorry, no. Hand-waving is talking about how NOW there are problems but TrainerRoad COULD do something different. And that the majority of people who come to TR and rely on the ramp test, or AIFTPD, SHOULD have known by checking the Internet to knock down their FTP a few points because sweet-spot work felt too hard.

How the hell are they supposed to know what too hard is if they’re inexperienced, time-crunched riders who bought a service and fully trusted it to have everything dialed in and spot on (because of ignorance and inexperience and not knowing just how hard and delicate that is to program for a mass-market product, not because TR failed somehow)?

Oh right, check “the interwebs” for enlightenment! Brilliant. Because there’s definitely no conflicting, vague, or contradictory information out there ANYWHERE about what FTP even is, much less how to set it or use it, right? And all of that perfect and wonderful information also tells you how hard things should feel, because RPE is totally precise and easy to communicate to everyone, ESPECIALLY inexperienced newbies.

Nah… just enlighten yourself on the Internet, try the ramp test, knock off a few percent (why? you think EVERYONE gets overestimated?), and go knock off a 2x20 to start if you’re “a complete novice”. That’s your coaching advice!?

And I’m “hand-waving to justify kinda mediocre training principles”. Suuuuuuure.


Not sure what you’re referring to… can you clarify, please?

I wasn’t pinning anything on TR. The ramp test overestimates the FTP of those with strong anaerobic responses, that’s a pretty well-known reality. It’s a great tool for probably 2/3 of the people, and maybe 15-20% (estimated) end up with their FTP set too high. For those people, sweet-spot feels like threshold and it’s very hard to progress time in zone.

I’m one of those people, and I fought that battle for a good while. I have 1400W peak power, but barely a 200W-ish FTP. And for my first 18 months of using TR, I had no idea that I was strong anaerobically, nor that the ramp test could overestimate for someone, etc. It was really, really hard to make progress.

But yes, to your point: TR has changed and evolved. Workout levels, AIFTPD, adaptive training, all these are tools to try to make the training system smarter and more adaptable to a population with such huge variation in capabilities and experience. I think they’ve done great work, even though I also believe they have a ways to go still.

Not sure if that’s what you meant, though?


I have given a sweet spot plan and basic advice to at least 3 untrained people who are part of another cycling community and all of them have progressed from 2x20 to 1x60 in a matter of a few weeks. So yes, that is my coaching advice!

1 Like

I’m glad it worked out well for those three.

I’ll note that your response has nothing to do with the ramp test, potential overestimation, or TrainerRoad’s PL’s and therefore is not relevant to this thread. But honestly, I also have no further interest in arguing with you. God bless, and go do whatever you feel is best.


Who says I am. I have used it in the past, I don’t any longer. But I learn a lot by being here so I still am on this board

The ramp test is inaccurate for far larger of a percentage than that. I have yet to see a single person in quite a large sample perform anywhere near the “ftp” that the ramp test spits out.

Easy when you give them a nice low FTP number :face_with_hand_over_mouth: :laughing:

1 Like

Is it though? Your anecdater and internet bias proves nothing :slight_smile:


I think AI FTP has adapted in recent months to move away from matching the ramp test. This was alluded to by Jonathan in podcast #466 (or thereabouts) saying that AI FTP was giving him a lower number than previously but he was getting more productive training.

It also explicitly now is influenced by your recent FTP history (I can’t find the reference, it was tacked onto a historic announcement thread) and your workout surveys AI FTP Detection Update - #379 by SarahLaverty

The impact for me has been significantly higher Progression Levels in SS and Threshold than previous years before getting an FTP bump. A little bit skewed by me also self-selecting alternates to extend TTE rather than accepting intensity increases.

3.5 minute intervals in SS are worthless


You mean using different operational definitions of an arbitrary measure produces different results? Who could have seen that coming?

The value in any FTP test is the extent to which it can be used to structure training. If you’re using a training system based on a particular test methodology, and workouts are appropriately scheduled, the test is doing what it needs to.


The ramp test is just like using 220 minus age for max hr.

1 Like

Performance is the performance itself and not the ftp number generated


How do you figure that? :thinking:

If you don’t have a number, how can you asses a timed effort?