More than happy to cooperate. Let’s get a few things out of the way first: I’m not an isolated case, please take a look at @Rizzi ‘s post:
Secondly, here’s a little poll with the training approaches of some of the people in this thread that might be concerned:
Doesn’t look like we’re “asking for it” due to aggressiveness of approach.
Lastly, I know full well that I can manually set my FTP to what I believe is a realistic value and get good training. That’s how I progressed last year (not manual, but I find my old AIFTP to be realistic), that’s how I got a good week in the middle of January. If anything, this shows that the new AIFTP I’ve been given is suboptimal for my training plan.
Now, about my training block this January. Here’s more detailed context: I started the year with an FTP of 346 from 2025, having finished a training block and due a detection, expecting increase. Maybe a few watts, as per usual, as I’ve been on a steady, sustainable rise through 2025 with no signs of slowing down. The AIFTP in beta gave me 381 (+35 watts!!!) and my first week was a disaster. That’s all good as it was beta and I was testing. Nate and Sean looked into my case and suggested that I use an FTP of 345 which was produced by a manual run of AIFTP detection from them, just to get me back on track. It did so, and I rated Scylla (Thr 4.7) Hard and Galena (SS7.2) Moderate with it, which made me think that I am indeed back on track and ready for the increased FTP I was expecting after the 346, so I increased my FTP to 353 manually. This isn’t a random value - it was produced by one of the detections Nate ran when looking at my profile, so let’s take it as baseline for my physiological zones, although I think it may still be a few watts optimistic. Nevertheless, here are the corresponding zones:
Sweet Spot @353: 311 - 332
Threshod @353: 333 - 371
VO2 @353: 372 - 424
Then came the release of AIFTP bumping me up to 366 and I started struggling again. Let’s look at watts to find the reason:
First workout @366 was Galena-1: 20@329 followed by 20@344, rated Hard. That’s one interval at the high end of my sweet spot and one in lower threshold. Not short, too. 20 minutes is respectable duration for continuous work. Looking at watts, this is more of a threshold workout than it is sweet spot, but it is mislabelled due to the FTP inflation. It is within my capabilities as an athlete, but it does not align with my plan and goals. Also builds more fatigue than necessary (and than expected). Lastly, for reference, the Galena I had done the week prior had 20@324 watts as its hardest interval, which is a whopping 20 watts less than the SS workout that came just 4 days later.
Second workout @366, Reinstein-2: 3x12 Threshold where each 12-minute block consists of 4 OUs: 2@337 + 1@373. The unders are fine, but the overs are in VO2 land. Ultimately, it was me losing my breath again and again that caused me to consider the workout a failure. It was not about accumulating and managing burn, but about VO2 capacity after the first set. Maybe I could’ve pushed through without backpedalling on a better day, but that still wouldn’t be an OU as it is meant to be. For comparison’s sake, I had an OU with identical structure the week before - Scylla (hard). It had unders at 335W and overs at 355W. So Reinstein-2 came with basically the same unders but overs at 18W higher. That’s not smooth or natural progression for a balanced approach in threshold zone.
I’ve been pushing my next Sweet Spot workout (Tray Mountain-1: 3x15@329, so top end sweet spot) forward in time for the past 2 days because I’m really not looking forward to it. The algo downgraded it to fit within my capabilities and I don’t expect problems completing it, but I’m frustrated and I still have slight soreness in my legs from the failed Reinstein-2.
I’m concerned that this is happening to other users as well. They might be fitter or tougher than me and getting through workouts instead of failing, but that doesn’t mean that their workouts are correct and serve their intended purpose. I’m also concerned that this experience is a recurrent one in my case. The first time it happened may have been in beta, but it still happened, and there’s nothing indicating that it won’t repeat on my next AIFTP detection.
Last observation: the model working with the new AIFTP is pushing me towards all time PRs on a regular basis (Reinstein-2 matched 12m, Galena-1 had PRs in the 15-20m range, Berge in the 60s-90s range) in my first base block already. I see that as a huge red flag.
Looking forward to making the product better for everyone!
P.S. If anybody has examples with watts like the two I gave (Galena@345 → Galena-1@366 with the 20W bump over 20min and Scylla@345 → Reinstein-2@366 with the 18W bump in overs), please share. This is the hard evidence that will actually make a case.