Hi all,
With the updated API Policy (June 1, 2026), Section 5.3 prohibits using Strava Data "in connection with the development, training, evaluation, or operation of any AI Application," explicitly including "ingestion into a context window or working memory."
I'm trying to understand how broadly this applies. My app (The Personal Shift, Client ID 191370) does the following:
Fetches athlete A's own activity data via the API
Passes it transiently to Google Gemini (inference mode only — no training, no fine-tuning, no embeddings, no persistent index)
Returns a personalized coaching note displayed exclusively to athlete A
The Strava data is not retained in the AI layer beyond the 7-day cache
This is strictly one-to-one: athlete A's data is only ever used to generate output for athlete A. No aggregation, no third-party data sharing, no model training of any kind.
My question: Does Section 5.3 prohibit this type of use, or is it primarily aimed at bulk data ingestion, model training, and aggregation across athletes?
I emailed developers@strava.com but received a generic response. Hoping someone here — or from the Strava team — has dealt with a similar use case or can offer clarity.
Thanks
Answered
is my app compliant?
Best answer by Mark_AJ
Hey! I appreciate the detailed breakdown, but Section 5.3 answers your question directly. Strava explicitly includes "ingestion into a context window or working memory" as a prohibited use. That's precisely what you're describing. Transient, inference-only, one-to-one, 7-day cache or not, passing Strava data into Gemini's context window is the exact use case the policy calls out by name.
The intent of your feature doesn't change what it technically does. Your app will not pass review with this in place.
Login to the community
No account yet? Create an account
Enter your E-mail address. We'll send you an e-mail with instructions to reset your password.
