Is there a plan to increase the AI functionality to include daily feedback on sleep, resting heart rate hrv and general exercise perceived exertion?
How would that work? Xert already has three tiredness statuses (low, high, peak) which is based on the training load, and not a measurement of how the body responds. Should it basically consider you as tired or very tired if some of these values are too high/low?
Hans similar to what training peaks uses with an AI communication like fascat…a combination of the user daily input with the AI communication directed by the science of XERT…Everyone has a great function I’m hopeful for XERT to lead the way as the science is far better and accurate than other platforms. Now that XERT has a racers function there is a for greater application and population that would benefit from XERT’s knowledge.
1 Like
There is a big hype these days about AI. It is not
magic. You still have to do a lot of development to make a good system.
But some kind of integration between Garmin metrics like Sleep Score, Recovery Time, HRV Status, Acute Load, Recent Sleep Score, Recent Stress, and Training Readiness can make sense. Xert does not know if you are very tired, sick, etc. It does not include any physiological measurements, only the power output in the calculations.
But in order to do this, I think you need to start with a few typical cases where the current model falls short, and see what kind of addition measurement could help and how that would change the advice.
1 Like
“There is a big hype these days about AI. It is not
magic. You still have to do a lot of development to make a good system.”
100% agree with this statement. Poorly executed “AI” has been implemented in a lot of places where is makes things less functional. Duolingo is an app that comes to mind where “AI” implementation has been poorly done
I much prefer a go slow and get it right approach.
1 Like
Yes, totally agreed. Was listening to a podcast with Paul Larsen on what Athletica are doing with GPT in their science. A GPT trained on Xert’s documentation and key literature would be ideal for coaching advice.
Yes XERT has an idea of your low, high and peak buckets and can somewhat predict your freshness levels but the part I feel that XERT really lacks is that feedback side.
Yes XERT has a sliding scale. Is it actually that useful? Probably not because its not finding out why are you really adjusting that slider? If XERT was predicting correctly from all your training where you are at then the slider should not really need adjustment at all.
To me it would be good if on the planner you can put in some detail about how do I feel today. Technically I am in the blue and fresh but you know what my quads feel tight and my calves feel extremely tender to touch. Based on some feedback before a workout it should then adjust the suggested plan for today based on that.
Then on your ride data that gets saved again have the ability to some how capture some detail around how you felt during the session and after.
I think this is where the AI part starts to kick in as that can analyze your data plus your comments and make on the fly adjustments to todays plan and start to understand how you as an individual actually function.