How AI will transform Strava


The social network for athletes is undergoing its technological revolution. At the heart of the changes are new features linked to artificial intelligence and machine learning, tools that can radically change the face and way of using the platform.

This week the annual event took place Strava in Los Angeles, “The Camp”. For three years now, this has been the moment when the most popular social network among athletes has announced its most important new features. As usual, the platform took the opportunity to announce new features, some of which, such as “dark mode”, had been expected for many months. There are therefore several new possibilities in the way of displaying activities, in the type of data provided to the user or more generally on how to navigate the application.

But during this 2024 edition, Strava also moved forward into another area, vizartificial intelligence and machine learning. Although there is nothing original about this movement (not a technical conference has taken place in the past year without the word AI being uttered repeatedly), it involves, for the platform, significant stakes. Given the amount of data generated on Strava (ten thousand billion activities downloaded by more than 125 million members) and the predominant place that the network occupies in the large community of athletes, it is appropriate to look at what Strava wants to do AI and how its algorithm will evolve to meet the new challenges posed by AI.

As Matt Salazar, the new CPO (Product Manager) of Strava, whom we were able to interview, explains, “we are only at the beginning of this integration and the changes it implies are significant”.

One AI, three main applications

As a reminder, the integration of an AI layer in Strava happens at several levels of the platform. The first of these concerns route generation. This was already possible on previous versions of the application, but in a rather basic version. Strava was able to locate its user, identify the most used segments in their geographic area and imagine a route that could correspond to them. AI will of course allow us to go further. On the one hand, the user will be able to more easily determine a starting point by “pointing” to a map. From then on, the route generation system will analyze more data, especially those linked to the user’s habits to offer him more routes generated with more information, and especially those coming from “activity maps”, in other words based on the most used paths depending of the time of day. As a result, route suggestions should become more personalized and interesting.

© Strava

The other aspect on which the integration of AI will have major effects is what the platform calls “data integrity”. This is just a matter of checking the relevance of the data recorded by the users. In fact, as with certain video games, Strava relies on a few cheaters who record fake activities to achieve the famous KOM (King of Mountain), which rewards the best times on each segment. With its new feature, Strava will now be able to detect suspicious activities and revoke them. At the moment, it is already a matter of verifying the relevance of all KOMs, but Strava will be able to rely on machine learning to go even further, that is, to judge the veracity of an athlete’s activity compared to his past performances. At this point, Strava engineers know they’re walking on eggshells. This is confirmed by Matt Salazar, who explains that it is about finding “the right dose and not giving the impression of punishing the user”. Initially Strava will therefore only query certain performances in the hope that this will be a sufficient deterrent.

Read : Strava has developed a deadly weapon to detect cheaters

The last aspect where AI will change the experience on Strava is also the most ambitious. This is the “Athlete Intelligence” feature. This aims to analyze each activity downloaded by a user to help them better understand their performance and progress. Thus, Strava is able to summarize the activity and contextualize it in relation to the subscriber’s previous performance. We were able to try out this feature for a few days, which is currently in beta, and the results are quite interesting. However, this is only the beginning according to the Strava product manager, who confirms that the functionality will develop quickly. “At Athlete Intelligence, we can say that we are on v1.0, but that we will very quickly move on to the equivalent of v3.0,” says the technical director. “We are currently testing several LLMs (large language models) and we have very encouraging results,” he confirms.

Strava Athlete Intelligence
© An example of an activity analyzed by Athlete Intelligence

In fact, the possibilities of such an application are enormous for a platform like Strava. More in-depth performance analysis, coaching and preparation for races with personal plans or even daily monitoring of activity, Strava has the opportunity to go in several directions and add some assets to its service.

The limits of AI without voice dictation

On the other hand, as ambitious as it is, Strava is in no rush. When we ask Matt Salazar about the benefit of adding voice dictation to the application, he tells us that there are other, more critical projects before considering such radical changes. The brand’s CPO confirms to us that Strava does not, at least not for now, work on such a new feature. However, this would make sense, especially in light of an increasingly important role being assigned to AI.

A strategic shift, a sign of a new direction

The fact that this Strava Camp 2024 was marked by such a technological shift is not insignificant. This event is also the first of a new management team, which has a very special DNA. Matt Salazar, whom we were able to interview, is a former member of Epic, the studio behind the game Fortnite. Even more emblematic is Michael Martin, who took over the reins of the company in January instead of the historic CEO Michael Horvath, coming directly from Google/Youtube.

Strava Camp 2024 3
© Dimitri Charitsis – 01net.com – The new CEO of Strava is a former Google employee

It is therefore not surprising that this new management team will first address the technical aspects of the platform and these key gaps. As such, the arrival of a dark mode is finally emblematic of Strava’s technological shift. The platform was often mocked for its delay on this point. Undoubtedly, the innovative leaders of the company wanted to make this dark state into a symbol. Moreover, the announcement of the functionality gave rise to a special sequence during Strava Camp 2024, as a “one more thing” sprinkled with self-deprecation.

Read : Strava is adding two highly anticipated features to its platform

This signal sent to users who had regularly complained about the absence of new functions and the ambitious announcements about the integration of AI in Strava confirms a major technological shift, but above all essential for a social network that wants to establish itself in the long term term. as the athletes’ digital companion.

🔴 To not miss any news from 01net, follow us Google News And WhatsApp.

Opera One - AI powered web browser
Opera One – AI powered web browser

By: Opera



Source link

Leave a Comment