The importance of commenting your sessions

[Marjaana]

Good morning team,

Thought I would get a conversation going about the importance of commenting your sessions.

Now, we all know we are busy. Our days are often comprised of ticking one box and onto the next thing. We finish our training session, hit save, and jump in the shower and forget about the session data.

And sometimes it just is like this. We do what we gotta do. I’m hoping to illustrate with an example of what happens when you give your coach IRL or AI coach some context to understand your situation.

Here’s a swim session I had on calendar this week.

Now, AI coach missed that I had Ironman Texas 3 days prior and had no idea why only did 1500m out of 2000 which was planned. AI coach gave me this feedback:

I wasn’t overly happy with being bitch slapped for taking it easy
So I gave AI coach some context. If you’ve hang around with us for a while you know we are all about context.

And here’s the instant feedback I got:

Pretty cool eih?!

My point is: whether you have a AI coach or a real life person coach - it is impossible to make up reasons why sessions get altered.

Use AI coach ! You can even ask questions about training , or share some thoughts you had during your sessions with us (AI coach and/or real life coach). Heck, I probably share too much information sometimes. I even shared my poem I made up for IMTX during a particularly fun run session.

Have fun with your partner in crime, AI Coach. You’ll be surprised what you find out!

Here’s my post session routine that I do after every session.
Stop watch, open Garmin, wait until activity syncs, open Athletica (save a shortcut to your phone for quick access), open session, glance over the data to make sure it reads correct, then
RPE (we spoke about RPE on The Athlete’s Compass episode ‎The Athlete's Compass: Beyond the Numbers: RPE and the Art of Listening to Your Body on Apple Podcasts
I also choose how I felt during the session and finally a few words about the session.
Hit save and wait few seconds for the feedback.
In the meantime I’m stretching - even before I get back inside the house and to the chaos of my every day life.

I do this right after the sessions because my thoughts are fresh in memory and if I have something important to tell my coach he has it there when he sees over my training for the day. If I leave it lingering, I will never get back to it.

Do you leave subjective feedback? Why, why not?

What are the funniest or surprising things your AI coach has told you?

IMO this feature is pretty incredible and I have a feeling none of us are using it to its fully capacity … please share how you are using it and what you‘ve found useful.

… can’t wait to hear what ya’ll share with me!

MJ

[Ellowinowh]

I’m a fresh user and just subscribed to a paid membership and to be honest I didnt even know this “feedback” was really useful, thought it was just for myself write something about the session, not a clue that the AI will take consideration of it!

[Marjaana]

@Ellowinowh welcome to Athletica! I highly encourage you to experiment with it

[Ellowinowh]

Will do, definitely !

[Ellowinowh]

Haha love it !
Had to stop at 1h15 training instead of 1h30 and told the coach about it!

[Marjaana]

Isn’t it awesome? ! thanks for sharing this such a good example of “life’s surprises”

[nmc]

This is really helpful! I would love to see more examples of how this is incorporated

[headgeek]

the main reason I prefer AI coaches over humans is they react and do something (adjust the training for instance) in every run.

I don’t expect a human coach to act up to a random easy run.

One day my AI overlord will yell at me through my watch (or my brain implant) : “push it you squishy lazy human!”

[mpderksen]

curious if my comments will effect programming, or just the comments. I have the comments turned off. But if I said, “I really like tempo runs on Wednesday”, would that result in tempo runs moving from Tuesday? I just assumed, for example, that only my RPE and Thumbs up/down would effect the length of my Sunday long run. If I start mentioning fatigue, and no change to the RPE, would the AI pull my long run down from 90 min to 70?
Curious what effect on PROGRAMMING my comments would make. Not just on the feedback text.

[Marjaana]

Hey
At the moment I believe AI Coach summary does not touch the plan, perhaps @Prof or @Andrea can confirm?

[Prof]

Not yet @mpderksen. But we in fact just had an update to the quality of the advice being put out by AI coach so could be worth considering now. What you are suggesting is of course in the pipeline.

[Andrea]

@mpderksen you nailed it, yes. That would be great indeed, as @Prof mentioned.
We had long internal conversations about how it would be like. At the very moment, the influence of the text input on the programming is limited and balanced by the loads of the previous sessions, and subjective feeling … you cannot “steer the boat” with a single message. Eventually this is something that we would like to have, but the discussion around how/when/what is taking some time. On the surface it might appear quite easy, but the edge cases and exceptions are a lot and might be difficult to handle.

[SoCal1x]

Fair to call this system an optimization (please challenge that thinking if I’m off)? We often find in finance that users of optimization platforms (generally speaking, regardless the form of implementation) tend to “steer” the output to the desired result (that is…the optimization is overridden by qualitative factors/heuristics). This, as opposed to simply trusting the math (we can argue whether any such math is relevant/should be trusted anyway, but that’s for another time…). But in this case, we’d rather want a qualitative input (that perhaps can be quantified with RPE, etc.) to effect a change in plan in the near-term to a modest/moderate extent, while maintaining the longer-term plan. I find such challenges fascinating (esp. in the context of how we understand AI to work) and would love to hear how you are working through the process. Podcast topic (without giving up too much of the secret sauce!)?

[Marjaana]

Just thinking out load here, @SoCal1x … are you suggesting that if a user says: hey I really enjoy my wednesday vo2max sessions , can I do them twice a week for the next 6 weeks? then the system would plot in vo2max sessions twice a week for 6 weeks, regardless of what the long term plan was originally?

[SoCal1x]

Maybe that’s Level 2? I was thinking at first perhaps something like this:
Considering I raced yesterday, I might suggest this morning that my legs are dead and that I might want something lighter today than the Vo2 session (not total recovery, but maybe Z2). So I might suggest I want an RPE of 3-5. So the platform might rework the order of the week, or otherwise shift today’s workout to something lighter, but still on plan.

[Phil]

A few weeks back, I had a suggestion to @Prof about sequence of sessions in a recovery week. In other words after a race / event.

For med volume cycling I now see

Recovery
Aerobic
Strength Endurance
VO2 Max

My view being that that makes for a better recovery week progression than higher intensity earlier on. It certainly fits the timeline of how my physical recovery and fatigue often looks.

[MartinM]

I like this feature but I find the coach can be a bit too forgiving. Can we add desired sentiment (like “I bailed early to go have beers with friends and I don’t want you to sugarcoat the feedback…l”) in the content to get something more like my mid 90s swim coach would say?

[Marjaana]

Omg I’m loving this! You like some tough love don’t you @MartinM ?!

[Andrea]

Very interesting thoughts. Taking notes here.
Our favourite analogy is that of the self driving car, where you have a higher-goal of going from point A to point B, and then intermediate goals or tasks or constraints, related to the environment, to the geography of the course, to the fuel usage, etc.
It’s not that far from what Athletica does indeed, it’s a similar hierarchical structure. Of the many ways you have to get to Rome (race A), the logic finds a suitable one, by respecting the constraints we are providing (time, modality, load, etc). It’s not a huge lift to introduce ways to make the car steer locally if we wanted, but it comes at the cost of having many unintended effects, let alone the fact that the scientific literature that can support the modelling choices is quite limited in these regards. So we are trying to make the transition happen, slowly but steadily

[MartinM]

I just tested something and wanted to share it here. When you get one of the infuriating “Your actual load was x, which is way below/above the y that it was supposed to be,” you can include in the comment a question about what it should have been and get feedback.

For example, I had an easy bike ride today, heart rate target fed through to the Garmin, and while I’m 99% certain the issue was my power meter glitching, I revised my comment to include a question about what I could have done to increase the load, and it came back with the appropriate power target (which I will in the future check before leaving the house…)