AI workout planner - how to be compliant

You can get an athlete’s activities via GET /athlete/activities (it also has a before/after range you can set). Each activity has an id which connects it to streams, kudos, comments, zones etc. Iirc this returns summaries, you can get more details for each activity via the GET /activities/{id} endpoint and then the same id can be used for the activities/{id} endpoints, or the segments/segment efforts and so on.

Note that “GET /athlete/activities” is paginated, so you’re gonna have to keep asking for pages, until you get no results (unfortunately they don’t tell you how many pages you’re gonna have to request, or how many activities there are).

To my understanding you shouldn’t need more than the summaries, unless you want to grab more data for the AI plan in a way that it factors in performance, stats etc. Then you need the segments, effors, zones and whatnot.

You could check the swagger playground for more, but you can also use Postman (it’s a bit finicky to set up, but can be fully integrated and since you can get the swagger json from the playground, you can also auto-generate the full API, if interested I can walk you through it).

Thanks for the reply!

I have successfully pulled the activities and everything is largely working, I was more concerned about Strava’s guidelines about training AIs from user data, and how this is interpreted.

Based on their showcasing of a similar app ‘Roast my Strava’ which takes user workouts and feeds them to an LLM, I’d assume my app wouldn’t have issues. I’m just not sure if I need to be aware of anything regarding storage of workouts, what exactly is given to the LLM, etc. in order to comply with Strava rules.

I don’t think the opinions of other users will benefit you much, in the end Strava decides how their rules have to be interpreted. So you should ask developers@strava.com about it.

I replied before to take a look at the API agreement and the Brand Guidelines, but I guess it’s being reviewed...for what reason I have no clue, both links being strava’s.