Artificial Intelligence and Sports Apps

Recently Strava added MCP AI decorations, with the idea of making workout feedback more interesting and dynamic. In practice when Garmin Suunto and Runna did that they forgot a key aspect. Allowing humans to add real context.

AI is a guessing algorithm, not true intelligence. It will look at data, but it won’t understand it. It will only correlate it to other data and provide a guess according to other contextual information.

If you go for hill training and it wants your pace, it doesn’t consider the elevation, heart rate, region of the workout and more. Runna is a premium app that doesn’t look at landscape and suggest a good training route, to maximise the gradient and training effect it would like you to get.

If you find your own real hills it will tell you off for not having the right pace, despite you sustainably running in the vineyards above Luins.

It also doesn’t recognise multi-sport training. Runna wants you to be a runner. Strava wants you to workout every day, as does Apple, but they don’t look at your fatigue, form, and fitness. These apps give instructions but they don’t take form into account and that is limiting.

I run, I hike, and I cycle, and by doing all three sports my cardio might show that I’m optimal, but my leg muscles are shot. They’re tired from running, so if I try to keep up with cyclists on optimised bikes, I don’t. I exhaust my legs, and myself.

In my eyes AI should run a “full system diagnostic”. It shouldn’t look at sleep as a single block, but rather as a series of naps, for people above a certain age. I very often close my eyes, dream, and then wake up a first time, and then I go back to sleep. That REM sleep is ignored by Coros, Apple, Suunto and Coros. Coros wants a solid three hour block of sleep, to qualify it as sleep. It ignores the “naps” or polyphasic sleep, to use another term.

When I attemot to converse with LLMs I see that they don’t have a clue about context. They see data, and they try to match tokens to tokens, without actually understanding it like a running coach, or a cycling coach would.

Claude, Gemini, Euria and MyAI to name just a few are not interesting to use for sports analysis because they’re prohibitively expensive. 20+ CHF per month, for guesswork, is not interesting.

In my eyes, and in my experience, AI is being sold as a premium feature, and yet, from what I have seen, it is not ready for prime time. Gemini, MyAI by Swisscom, Euria and one or two others might provide some insight and some feedback, but if you dig deep you see that the feedback is skin deep and superficial, rather than in-depth understanding.

While writing this blog post I experimented with connecting Claude to Strava and it said “method not allowed” so we are still in the infancy of this opportunity.

Paradoxically one of the reasons Coros is interesting over Garmin, Apple and Suunto is that it looks at effort, calculates rest needed, and gives a number. It doesn’t micro-manage you based on flawed sleep data.

Artificial Intelligence at the moment, is not ready to be of real value for the moment.