Spotify’s AI-agent strategy

Spotify’s third quarter was strong: the service reached 788 million monthly active users and generated nearly €5 billion in revenue. But on its earnings call, the company also made clear that it sees a bigger opportunity in how people may listen in the future. Co-CEO Gustav Söderström believes entertainment can thrive in an agent-led era, because it is not simply a chore people want AI to handle. As he put it, “Nobody asks an agent to ‘summarize a song’ for them.”

Instead, agents could help listeners find and enjoy entertainment in new ways, while Spotify aims to become the media service people can use through any agent. The company says it is already seeing good adoption inside these tools. Spotify was the first music service to integrate with Meta’s Muse assistant, and it also works with ChatGPT and Claude. Söderström described the strategy as turning agent growth into a tailwind rather than a headwind: Spotify is “leaning in and participating in the growth” instead of standing aside. Its bet is that conversational access can complement its own app, allowing listeners to discover and interact with Spotify through agents as well as directly.

How listeners could use Spotify through agents

Instead of opening Spotify, searching for an artist, and building a playlist, listeners could describe what they want in a conversation with an AI assistant. They might ask for music to accompany a run, a podcast about a topic they have been discussing with an AI, or an audiobook recap that resumes exactly where they left off.

Spotify wants to be the service agents turn to for those requests, making it possible to discover music and interact with Spotify through conversation rather than only through its own app. The company was the first music service to integrate with Meta’s Muse assistant, and it also integrates with ChatGPT and Claude.

Spotify says its goal is to be a media service people can use from any agent. Co-CEO Gustav Söderström said the company is seeing good adoption of Spotify inside these agents and is participating in their growth. The idea is not simply to have an assistant pick something to play: Spotify describes its service as something listeners can talk to, ask questions of, and instruct, whether they use it directly in the Spotify app or through another agent.

Personalized listening powered by Spotify

Spotify’s vision goes beyond making entertainment easier to find: it wants to make listening more personal. Co-CEO Gustav Söderström describes Spotify as an intelligent media service that people can use through its app or via another agent. Listeners can talk to it, ask questions, and give instructions, rather than only clicking through the app.

One example starts with a listener’s history. They could ask Spotify to review songs they have listened to, going back to 2008, and pick out their true favorites. Spotify could then speed those tracks up or slow them down, sort them by key and beats per minute, and mix them into a running playlist matched to the listener’s exact cadence.

That kind of personalization draws on Spotify’s years of listening data: the artists a person returns to, the songs they skip, and the music they discover. Söderström says Spotify uses its own systems to reason about a request and decide what to play, rather than sending listener data to partners. The result, in Spotify’s vision, is a playlist informed by an individual’s listening history, not a generic selection from an assistant that knows little about their taste.

Access, competition, and listener data

Opening services to outside agents can create tensions over who gets access and under what terms. Amazon’s decision to block Meta’s Muse from shopping on its platform is one example of how questions about access, security, privacy, and commercial control may lead major players to restrict outside agents. The dispute points to a broader challenge for services that want people to reach them through third-party assistants: participation may depend on the platform being willing to allow an agent in.

Spotify has said it wants to be available through any agent, and it was the first music service to integrate with Meta’s Muse. It also integrates with ChatGPT and Claude. But the company says that working with agents does not mean handing partners listener data. Instead, Spotify says an agent asks it to do something, and Spotify processes the request with its own systems to decide what to play, without sending listener data to partners.

That distinction matters because Spotify can draw on years of listening history—including artists people return to, songs they skip, and music they discover—to shape recommendations. The company’s approach aims to make agent-based listening more personal while keeping that data within Spotify’s own systems.