Garmin Training Readiness vs. AI Endurance: Which do you trust?

Questions
1

(Quick disclaimer: If you’re just here to comment “readiness data is useless, just listen to your body,” please save it for another thread. Looking for a data-driven discussion)

For those of you who use both the Garmin ecosystem and AI Endurance to make data-driven decisions on when and how to train: Do you rely more on Garmin’s Training Readiness or AI Endurance’s Aerobic Recovery?

Here is my current dilemma:

AIE lean heavily on RHR and HRV. For me, RHR is often all over the place. My true “rested” RHR is around 46. Often, I’ll take two rest days, my RHR will sit at 52, and AIE will tell me not to train. If I decide to ignore it and train anyway, my RHR drops to 48 the next day and the algorithm(chat) ties itself in knots trying to justify it.

Garmin’s Training Readiness factors in a wider net of metrics and doesn’t anchor as strictly to daily RHR fluctuations (mostly just what is incorporated into the sleep score).

I’m really tempted to just follow the Garmin score for the next few months and live with AIE’s nagging notifications about canceling my workouts.

Example:

I’ve been traveling for the last three weeks. My routine isn’t exactly what it is at home, but my habits haven’t gone off the rails. I’m sleeping 8–9 hours a night, hitting my protein targets, keeping fat low, and avoiding alcohol.

Despite feeling completely rested, my HRV has been slightly off and my RHR is elevated (50-55). AIE has been aggressively canceling all my workouts, while Garmin is giving me the green light. Given how much I’m sleeping and felling rested, I can’t help but wonder if AIE is just being far too severe.

What are your experiences? Has anyone else found AIE too conservative compared to Garmin? I would also be really glad to get Mark’s input on this to hear his perspective. I have had lengthy discussions with the chat around that and I understand why it’s important for AIE.

2534
25341075×2393 283 KB

2535
25351075×2393 215 KB

(P.S. Apologies for the French in the Garmin screenshot, the Connect app doesn’t let you set a language independent of the system language)

2

I basically rely on the AI endurance prompt, my Garmin doesn’t have the training readiness score. The closest is the daily suggested workout which might recommend rest.

Generally, I do the workout and ignore short-term fluctuations even if RHR je high and HRV low as it doesn’t know if I’m having stress at work, travelling, meal timing, same/different bed, etc. and I’m not convinced Gemini is there “yet” to do this.

But if I get bad nights sleep after à hard workout with another hard workout that day, if I do it I’ll probably end up skipping the workout the day after if I get a third night of bad sleep, so I often accept the recommendation to convert the hard workout to an endurance one.

But if I don’t feel like it, motivation (etc), I’ll skip the workout even if the numbers are good.

3

Thanks, that make sense, that’s the kind of thing I wanted to read, what other athletes use as “ready to train”.

4

Hi Maxime,

I used AI since 10y now, including AIE and other plaforms in the past.

For me, garmin is optimistic or pessimistic (conservative). I prefer to maximize chance when i write the prompt each morning including my feelings, garmin score, night duration, and sore scores. AIE hits the target each time if i’m honest.

5

Interesting, tbh, I haven’t really used the check-in feature of AIE, I’ll certainly try it. Did AIE ever increase the intensity of a workout or tell you to trained on a rest day?

6

Hi Maxime

I would also say use the daily check-in as you can then have a conversation with AIE and tell it for example that you’re feeling good.

Since the daily check in auto-launched chat after you check in, I use it every day. You can then tease out these conflicts and resolve them.

I don’t trust the Garmin outputs very much at all.

Thanks

Grant

7

Sorry for the late reply, currently on holidays :smiling_face_with_sunglasses:

It’s an interesting question and I won’t go out and say AIE is definitely right and Garmin is definitely wrong on the RHR recovery impact interpretation. The answer is as so often it depends: we’ve had a lot of feedback and iterations over the HR(V) recovery modelling and I believe for most people it is appropriate with the caveat that we’ll always err a little bit more on the side of caution (it’s much ‘cheaper’ to ignore a notification than dealing with the consequences of for example an ill advised hard training session)

It’s true that often/in most cases there’s more context than the bare numbers which is exactly why the checkin exists and is useful. So I think the checkin actually addesses that gap between “just trust the bare numbers/algorithm/modelling” and “just listen to your body”.