Designing Agentive Technology: Part 2
Over the past months Natalie Jensen has joined us as an intern in the San Francisco Labs. In her last piece Natalie wrote about a meetup she attended, where Chris Noessel spoke about agentive technology and its uses. In this piece, she analyzes a specific use case to further explain the different elements of agentive technology.
To provide a more in depth example of agentive technology, let’s look at Spotify. If you’re an Apple Music user or just not familiar with the platform, one of the main features that Spotify is known for are their playlists; more specifically the individually curated “Made For You” playlists. Everyday Spotify users get six new “Daily Mixes” all catered to different micro-genres of music with songs you frequently listen to, as well as a “Release Radar” playlist every Friday, and a “Discover Weekly” playlist every Monday. These playlists can all be considered “agentive technology”, as Spotify creates them for you based on your listening history. Instead of having to make a playlist of songs you like, Spotify does it for you; instead of looking for new music, Spotify presents you with 30 new songs every week. The Discover Weekly playlist is widely acclaimed for being uncannily good, and often mentioned in articles about what makes Spotify superior to Apple Music – so how does Spotify do it?
Spotify has utilized machine learning to build an algorithm that caters to each user based on their individual preferences. It’s a lot more complex than what I’m about to describe, but it mainly comes down to the playlists you listen to and your ‘taste profile’ made up of microgenres you repeatedly listen to. They take into account what you haven’t listened to from playlists with similar songs to the ones you frequent, as well as the categories those songs fall into and voila, a playlist “Made for You”. These playlists have been so successful other companies have tried to replicate them by building their own versions, but Spotify still has proven itself to be supreme with the user base of any other music streaming platform.
Going back to the original principles of agentive technology – the user needs to be able to start, monitor, tune, and stop the agent. The “agent” being Spotify’s algorithm that makes the various playlists for you. The start is simply creating a profile and listening to music. You’d monitor this by going on the app and listening to the playlists it creates for you, and if you don’t like what you’re getting you can tune these playlists by searching for new music yourself. The stop would simply be deleting your Spotify account. Remembering that the goal of agentive technology is disengagement, you can see how Spotify achieves by allowing the user to stop spending time on making playlists and searching for new music, instead presenting it to them and allowing the user to just click play.
This specific type of agentive technology can be referred to as a sort of “recommender system”. Amazon, Hulu, YouTube, Best Buy and many other companies have all created their own algorithms to recommend things to their users.
Whether it’s a show or a product, catering to consumers as individuals with these technologies has proven to increase company value and revenue. Making things “for you” is, and will continue to be, a growing trend – hopefully this gave you a little more insight about how “Made For You” is made.