Our first post on strive.ai! We’ve been focused on the engineering and data science of this application for the past 6 months and I’m excited to see it launched to a broader beta to test the features and infrastructure at a larger scale. So far we’ve had ~10 athletes using the application for about 8 weeks.
The initial beta yielded some great discoveries in terms of how many unique data situations would arise with a diverse set of users. Far and away the majority of energy and time invested in this app has been spent on conditioning the training data for the algorithms. Close second is getting everything up in AWS to make sure we’re prepared to scale out!
Probably the biggest change resulting from the early beta was to change from requiring the athlete to visit strive.ai to a model where the data is analyzed in the background and notifications are sent via email when something noteworthy happens. This is consistent with some of the other popular Strava API applications (e.g. relive.cc) and brings focus to events / discoveries that are valuable to the user instead of forcing them to visit yet another sports data site. I know I personally use Strava, Training Peaks and Garmin Connect so moving to more of a conversational interface is the right direction.
For now, we’re using email as the medium but I’m talking to developers@strava.com about the possibility of using comments or potentially their notifications as a broader strategy to enable conversational interfaces from their app ecosystem more broadly. Once the app has more users we’ll revisit this topic.
At any rate, look for a variety of posts pertaining to data science, endurance sports, technology and user experience as we move forward!