Spotify’s latest AI strategy is built around a simple idea: people may want help finding music, but they still want the entertainment itself. The company is positioning its service to work with AI agents, giving listeners another way to ask for music, podcasts and audiobooks that fit what they want to do.
Let agents handle the search
On Spotify’s earnings call, Co-CEO Gustav Söderström argued that entertainment differs from tasks people are happy to hand off. People may welcome an agent taking care of a chore, he said, but they are unlikely to ask it to summarize a song instead of enjoying it.
That distinction shapes Spotify’s opportunity. A listener could ask an assistant for music for a run, a podcast connected to a recent conversation, or an audiobook recap that resumes at the right point. The agent handles the request, while Spotify supplies the media.
In this model, discovery does not have to begin inside Spotify’s own app. A listener could describe what they want in a conversation with an AI assistant, and Spotify aims to be one of the services that assistant can call on.
Spotify is opening up to outside assistants
The company says it is integrating with several agents. Spotify was the first music service to integrate with Meta’s Muse assistant, and it also works with ChatGPT and Claude.
Söderström described the goal as making Spotify available from any agent. He said the company is seeing good adoption inside these tools and wants agent growth to support its business rather than draw listeners away from it.
That approach puts Spotify’s strategy in contrast with the friction emerging elsewhere. Amazon has blocked Meta’s Muse from shopping on its platform, showing that companies may limit outside agents when they have concerns about access, security, privacy or commercial control. The source does not say Spotify faces the same restriction, but the example points to a broader question for agent-enabled services: who gets to act on a user’s request, and under what conditions?
Personalization depends on Spotify’s own knowledge
Spotify’s pitch goes beyond making entertainment easier to locate. Söderström described the service as an intelligent media service that can be used directly in its app or through another agent. Listeners can interact with it by asking questions and giving instructions.
He offered a detailed example: someone could ask Spotify to review their listening history song by song back to 2008, select their true favorites, adjust the tempo, sort tracks by key and BPM, and combine them into a running playlist matched to their cadence.
The example illustrates what Spotify believes its accumulated listening history can contribute. The company has information about artists listeners return to, songs they skip and music they discover. That history can help tailor recommendations to an individual, where a general-purpose assistant may know less about that person’s tastes.
The value of the request is therefore not only that an agent can assemble a playlist. It is that Spotify can draw on a listener’s past activity to make the result feel specific to them.
Spotify says listener data stays with Spotify
Personalization raises a question about how information moves between Spotify and the assistants it connects with. Söderström said Spotify is not sending listener data to its partners.
Instead, an agent asks Spotify to do something, and Spotify uses its own systems to interpret the request and decide what to play. In practical terms, the outside assistant can provide a conversational route into the service, while Spotify says its own systems handle the listener’s data and the selection.
That arrangement is central to Spotify’s bet: agent conversations can bring people to its media without requiring it to hand over the listening history that makes its recommendations personal. Whether agents become a lasting route into entertainment will depend in part on how these partnerships develop and how listeners choose to use them.