Login required to started new threads

Login required to post replies

Prev Next
athletica AI?
Quote | Reply
hi everyone -
i learned about athletica AI today, when i heard that they'd signed ali brownlee as a sort of spokesman (or partner? or something?) can't find much about them online aside from their own website, though.

anyone tried it? would be curious to see how it compares with something like garmin's system (which i think is driven by FirstBeat).

i'm generally happy to build my own programs, like i have since i was a kid, but thought it might be neat to see if AI would come up with some very different ideas from my own. or whether it would be able to use my metrics more subtly than i can to adjust as i go and make the program more effective. on the other hand, would the AI know everything about my background, my life outside training, and would i need to stay with it for a long time for it to 'learn' how i react to different training, seasons, diet, etc?

curious to hear thoughts.

____________________________________
https://lshtm.academia.edu/MikeCallaghan

http://howtobeswiss.blogspot.ch/
Quote Reply
Re: athletica AI? [iron_mike] [ In reply to ]
Quote | Reply
Hi Mike,
As co-founder of Athletica.ai (full disclaimer), I was interested in Athletica reviews I might find on the internet. My search found your lonely message that nobody responded to (boo-hoo). I find that even interesting. Is it a fear thing, or what?
Anyway, here we are about 17 months later and I wondered - did you ever try it?
Alistair is indeed involved and contributes ideas to backend development with lots planned for the future. If you go to the 'ABOUT' section on the Athletica website you can watch an 8-min interview where Alistair explains why he joined us and what we are working on. Ultimately we are connecting leading sport science and coaching principles into code. In fact, less black-box AI and more coded logic with how professional high performance coaches and sport scientists operate. That means you can get a high level service at a low cost. We started building Athletica in 2015 and our BLOG page provides multiple examples of the principles that have been implemented.
To your point about needing to be on Athletica for a long time before 'it get's you' - yes - we were getting that a lot. But literally yesterday we started the implementation of the historical backfill for new users. That means that when you onboard, we reach back into your wearable history to provide you with immediate value (zones, power/speed profile, fitness, etc). As mentioned, it was only just implemented to no doubt there will be some growing pains with this feature as we are used to. But anyone reading and interested, give it whirl and let me know what you think...
Best,
Paul

https://athletica.ai/
https://hiitscience.com/
Quote Reply
Re: athletica AI? [plaursen] [ In reply to ]
Quote | Reply
Hello Paul,

thank you for engaging into the discussion.

Could you maybe share more about the product? Specifically:
1) what was the sample data used for training the model? specifically what are some of the key independent and dependent variables? in other words, what are you trying to maximise (were you looking at athletes results, perceived satisfaction or e.g. higher VO2max estimates) and what is found to be key contributor variables (potentially you wouldn't like to share those)?
2) you're mentioning the implemented SMEs model guidance being the key differentiation of the product. how much supervision have you included in the model and is this reflected in the data pre-processing to make the data guided for the model or is the model itself having classical logic rules implemented? or both?
3) can you share model performance assessment results?
4) apart from the above mentioned guided component to the model, what are other key differentiators to the market players like Humango or AI Endurance or Triq AI or any other?

Thanks,
Michal
Quote Reply
Re: athletica AI? [Michal_CH] [ In reply to ]
Quote | Reply
Hi Michal,

A pleasure.

1. Sample data are our users only at this point. Standard Banister (1971) TRIMP modelling with our own in house modifications.
2. Classic model logic rules (HIIT Science Chapter 8 and 9; Load and Load Response) implemented.
3. See testimonials. Happy campers.
4. Key differentiators include: Performance Profiling (OmniPower, CP + APR), Workout Wizard, Performance Reserve. See blogs for expansion on any of those items. Lots more on the way!

Best,
Paul

https://athletica.ai/
https://hiitscience.com/
Quote Reply
Re: athletica AI? [plaursen] [ In reply to ]
Quote | Reply
plaursen wrote:
Hi Mike,
As co-founder of Athletica.ai (full disclaimer), I was interested in Athletica reviews I might find on the internet. My search found your lonely message that nobody responded to (boo-hoo). I find that even interesting. Is it a fear thing, or what?
Anyway, here we are about 17 months later and I wondered - did you ever try it?
Alistair is indeed involved and contributes ideas to backend development with lots planned for the future. If you go to the 'ABOUT' section on the Athletica website you can watch an 8-min interview where Alistair explains why he joined us and what we are working on. Ultimately we are connecting leading sport science and coaching principles into code. In fact, less black-box AI and more coded logic with how professional high performance coaches and sport scientists operate. That means you can get a high level service at a low cost. We started building Athletica in 2015 and our BLOG page provides multiple examples of the principles that have been implemented.
To your point about needing to be on Athletica for a long time before 'it get's you' - yes - we were getting that a lot. But literally yesterday we started the implementation of the historical backfill for new users. That means that when you onboard, we reach back into your wearable history to provide you with immediate value (zones, power/speed profile, fitness, etc). As mentioned, it was only just implemented to no doubt there will be some growing pains with this feature as we are used to. But anyone reading and interested, give it whirl and let me know what you think...
Best,
Paul

hi paul -
thanks for looking this up. i am still sort of curious about the software and might be giving it a try soon. i'll keep you posted if so.

____________________________________
https://lshtm.academia.edu/MikeCallaghan

http://howtobeswiss.blogspot.ch/
Quote Reply
Re: athletica AI? [iron_mike] [ In reply to ]
Quote | Reply
I've been using it for about a year now, and it's been great. I mostly use it for a running program and just let my swim training flow into it to keep track of training load, etc.

They have been very responsive with questions and issues (most recently adding the option to push pace or HR to Garmin Connect after a whopping two people asked for it on the forum). Workouts adapt nicely based on your performance (or illness, or just slacking off) in recent days. I don't have a lot to compare it to, but it has been quite effective for me.

***
Quote Reply
Re: athletica AI? [M----n] [ In reply to ]
Quote | Reply
I have been using it for all my races this year and love it. I finished first in my AG in a 70.3 and in the top 10% at 70.3 world champs in Lahti.

Athletica plan includes multiple significant types of sessions for each discipline: recovery (after hard WE or races), HIIT (High-Intensity Interval Training)/VO2max, Threshold for speed, Strength (paddle for swim, low cadence for bike, hills for run), Z2/Fatmax/VLAmax. Each week also includes 2 strength sessions. When you are not ready or do not like a session, you can use the workout wizard, which offers other options to execute the intent.
Their AI is like having a personal coach. It tweaks your plan based on how hard you're going or if you're taking it a bit easy – just like having a top-notch coach in your corner. Plus, the workouts sync up with Garmin, and you can hop onto Zwift in a flash. The whole platform's constantly evolving; they're all ears for athlete input. You can chat about it on their forum; even the founder/scientist chimes in often!
Quote Reply
Re: athletica AI? [iron_mike] [ In reply to ]
Quote | Reply
This quote from the great Joe Friel in his latest Craft of Coaching issue could be worth a view...

Quote:
If your income is largely dependent on sales of templated training plans for a broad market, I suggest you look to make a change because this where AI can outperform you with lightning-fast personalization. To remain competitive going forward, every training plan needs to be highly personalized. This will cause some head-scratching because customization involves time, and time costs money. You’ll have to charge more.
Some big releases coming out soon so please drop on by if you're not afraid of taking time to learn new tricks...

https://athletica.ai/
https://hiitscience.com/
Quote Reply
Re: athletica AI? [plaursen] [ In reply to ]
Quote | Reply
How do the bike sessions work, do you have a platform that allows them to work directly in erg mode with a trainer kind of like trainerroad?
Quote Reply
Re: athletica AI? [lastlap] [ In reply to ]
Quote | Reply
I've been using the platform for the past few months and the bike workouts do not automatically show up like Training Peaks workouts do. I did find this video showing how to get a Garmin workout into Zwift though. I haven't tried it yet, but I'm still able to ride outside for a bit more...

So far, I've been happy with Athletica. I've been 1-on-1 coached, used cookie-cutter programs and just pieced it together on my own in the past. So far, I feel it's great bang-for-buck.
Quote Reply
Re: athletica AI? [AKCrafty] [ In reply to ]
Quote | Reply
So to do their bike workouts I also need to subscribe to a platform like Zwift....hmmm that's not great with the added cost.

With Traineroad they also have 'AI' adaptive training all in the one stop shop.
Quote Reply
Re: athletica AI? [lastlap] [ In reply to ]
Quote | Reply
lastlap wrote:
So to do their bike workouts I also need to subscribe to a platform like Zwift....hmmm that's not great with the added cost.

With Traineroad they also have 'AI' adaptive training all in the one stop shop.


To do the Athletica bike workouts, you can do them anywhere you like, ideally outdoors on the road pushed to a Garmin device (my personal preference), but also indoors with downloadable files for Zwift, Training Road and others.

We pride ourselves on being science-based in terms of our analytical tools and prescriptive logic. With great respect for TR as a platform for indoor riding, I'm not sure whether science is at the heart of its operation.

This blog describes a first-time competitor in the Kona Ironman and how Athletica enabled her a race achievement that went well beyond expectations. A user would use Athletica because they desired a similar outcome for themselves.

https://athletica.ai/
https://hiitscience.com/
Last edited by: plaursen: Oct 18, 23 13:29
Quote Reply
Re: athletica AI? [plaursen] [ In reply to ]
Quote | Reply
The blog link doesn't exist?
Quote Reply
Re: athletica AI? [lastlap] [ In reply to ]
Quote | Reply
Thank you... broken link... updated here.

https://athletica.ai/
https://hiitscience.com/
Quote Reply
Re: athletica AI? [plaursen] [ In reply to ]
Quote | Reply
I have used Trainer Road for a few years now as a recreational cyclist doing the odd Fondo. I plan on a 7 day road bike stage event (i will be completing as opposed to competing!!) from what i can tell you do not have a plan available for that?
Quote Reply
Re: athletica AI? [Zulu] [ In reply to ]
Quote | Reply
A few people have found that the program can be 'tricked' into events that they don't have preset plans for. You could set multiple B races for that seven day event, or multiple Bs and an A race for the day you want to target, etc.

If it doesn't work, you can delete the plan and try something else instead. (I've joked in the user group that when I'm trying to figure out a plan for something like a swimrun, the server costs go through the roof for a few hours.)

***
Quote Reply
Re: athletica AI? [M----n] [ In reply to ]
Quote | Reply
Seems there's no app available?
Quote Reply
Re: athletica AI? [plaursen] [ In reply to ]
Quote | Reply
plaursen wrote:


1. Sample data are our users only at this point. Standard Banister (1971) TRIMP modelling with our own in house modifications.
2. Classic model logic rules (HIIT Science Chapter 8 and 9; Load and Load Response) implemented.

I wish that people would quit using the word AI when they are really selling (possibly) ML or in this case what sounds like an algorithm driven product.
This said, I might give this shot. Is there a mobile app for android? I am assuming that I need to give this thing access to Garmin Connect? I am assuming that your PII handling is decent enough?

Next races on the schedule: none at the moment
Quote Reply
Re: athletica AI? [alex_korr] [ In reply to ]
Quote | Reply
alex_korr wrote:
I wish that people would quit using the word AI when they are really selling (possibly) ML or in this case what sounds like an algorithm driven product.

agreed.. but the rules of marketing aka late-stage capitalism, mean you have to have AI in the marketing materials for any new software product. We just had this discussion in our company.. not claiming AI means you are automatically several steps behind the competition.

I know Alan Couzens has been very happy with his implementation of ML for his systems, it does seem this could be a useful next step for coaches..
Quote Reply
Re: athletica AI? [alex_korr] [ In reply to ]
Quote | Reply
alex_korr wrote:
plaursen wrote:


1. Sample data are our users only at this point. Standard Banister (1971) TRIMP modelling with our own in house modifications.
2. Classic model logic rules (HIIT Science Chapter 8 and 9; Load and Load Response) implemented.


I wish that people would quit using the word AI when they are really selling (possibly) ML or in this case what sounds like an algorithm driven product.
This said, I might give this shot. Is there a mobile app for android? I am assuming that I need to give this thing access to Garmin Connect? I am assuming that your PII handling is decent enough?


Athletica uses a range of AI principles and algorithms, including numerical optimization, to provide intelligent solutions.


You do need to provide your training data through Garmin Connect or Strava. We prioritize the protection of your personal information and comply with GDPR and CCPA. During your free trial and subscription, we have robust security measures to safeguard your data, as detailed in our privacy policy. Upon account cancellation, all your personally identifying information is erased.

https://athletica.ai/
https://hiitscience.com/
Quote Reply
Re: athletica AI? [ThyTri] [ In reply to ]
Quote | Reply
ThyTri wrote:
Seems there's no app available?

Don't really need one, the sites works really well on your phone
Quote Reply
Re: athletica AI? [plaursen] [ In reply to ]
Quote | Reply
plaursen wrote:
This quote from the great Joe Friel in his latest Craft of Coaching issue could be worth a view...

Quote:
If your income is largely dependent on sales of templated training plans for a broad market, I suggest you look to make a change because this where AI can outperform you with lightning-fast personalization. To remain competitive going forward, every training plan needs to be highly personalized. This will cause some head-scratching because customization involves time, and time costs money.


Video walkthrough of latest UI/X update for coaches here.

https://athletica.ai/
https://hiitscience.com/
Quote Reply
Re: athletica AI? [plaursen] [ In reply to ]
Quote | Reply
I recently signed up for a free trial of TriDot, and after reading this thread I’m curious how the two platforms compare?
Has anyone tried both or switched to Athletica AI from TriDot?
Quote Reply
Re: athletica AI? [Biscuits] [ In reply to ]
Quote | Reply
Hey, I tried TriDot only shortly as the training plan it was giving me was not making any sense to me. Way too much Zone 3 work.

Also tried Humango, and not impressed. Way too much tempo work. For example, long runs backed by another run the day after. Running 2:45 with some tempo and then running another run the day after with more tempo. That is asking for trouble if you ask me. Why two tempo paced runs back to back? In fact out of 12 workouts, only one was zone 2, 2 strength workouts and rest had either threshold efforts, vo2max efforts or tempo efforts. Geez.

Athletica is well balanced in my opinion, following solid sports science principles allowing me to work at really high intensity some of the time (love their 30/30 VO2max) and most at low intensity. Of course build phase is more towards race pace as it should :D

In between these these three, Athletica clearly demonstrates the sport science that goes into its backend in its planning. As they say: train smarter, not harder.
Quote Reply
Re: athletica AI? [PushPlay] [ In reply to ]
Quote | Reply
Thank you for the info, I think I’ll give Athletica AI a try :)
Quote Reply

Prev Next