AI & Music

Your Next Playlist May Be Picked by an Agent

By the Starchild Music team · October 6, 2026 · 5 min read

A friendly abstract orb hovering over a turntable, sorting a fan of vinyl records

The short answer

On September 23, 2026, Spotify became the first music streaming service to connect to Meta's Muse AI agent, letting users play music and build playlists by conversation, and on September 29 OpenAI launched its own always-on agents, called Dots. Listening is starting to move into general-purpose assistants that act on a user's behalf, which changes how music gets discovered.

Two announcements in the last week of September show where music listening is heading. On September 23, 2026, Spotify announced that users can connect their accounts to Muse, the AI agent Meta released earlier in the month. Six days later, at its DevDay event on September 29, OpenAI launched Dots, its own persistent agents for ChatGPT subscribers.

We wrote last month about YouTube's Ask Music, which lets subscribers describe what they want to hear inside YouTube Music. The new development moves that conversation outside the streaming app. Music is now being controlled by general-purpose assistants that also manage a user's calendar, email and errands, and that can act without being asked each time.

What Spotify and Muse can do together

According to Spotify's announcement, once a user links an account, Muse can search for and play songs, podcasts and audiobooks, save tracks to the user's library, create and name playlists, schedule playback for set times (Spotify's example is a workout playlist at 7 a.m.), control playback on connected devices such as a kitchen speaker, and generate Spotify's Personal Podcasts from a user's briefings and notes. Spotify wrote that the agent "can play music on Spotify, save songs to your Library, build playlists, help you find a book or podcast, and more," all from a conversation.

Android Authority highlighted the cross-app angle: because Muse also connects to calendars and other services, it can, for example, see a road trip on a user's calendar and build a playlist for the destination. "Muse is trying to make listening another action inside a larger personal workflow," the outlet wrote.

Music Ally reported that Spotify is the first music streaming service to connect to Muse, after Ticketmaster became its first music-related partner earlier in September, and that the Muse app passed 2.8 million downloads in the US and Canada within 12 days of release. It also placed the move in a pattern: Spotify launched an app inside ChatGPT in October 2025 and a connector for Anthropic's Claude in April 2026.

OpenAI's Dots

OpenAI introduced Dots on September 29 as always-on personal agents that pursue goals in the background with limited supervision. TechCrunch reported that they run on the GPT-6 Astra model and are available to Pro and Business Premium subscribers. BGR and Yahoo Finance reported that Dots can connect to more than 4,000 apps through ChatGPT's plugin system. OpenAI's launch materials emphasized work tasks such as monitoring projects and handling invoices, and Yahoo Finance described the product as more enterprise-focused than Meta's consumer-oriented Muse.

Our analysis: what changes when agents choose

When a listener opens a streaming app, the service's own recommendation systems, editorial playlists and interface decide what is put in front of them. When a listener asks an agent to "play something for the drive," a second layer of software sits in between. The agent interprets the request, decides what to ask the streaming service for, and may never show the listener a home screen or a playlist cover.

That has a few likely consequences. Requests to agents tend to be descriptive ("upbeat, no explicit lyrics, something like what I played last week"), so the descriptive data attached to a track, such as genre, mood, tempo, credits and AI-disclosure labels, becomes more important to whether it gets picked. Context also matters more: an agent that knows a user's schedule can request music for a specific situation. And streaming services give up some control over the experience in exchange for being the default source when an assistant needs music.

Some important details are not public. Spotify's announcement does not say whether plays started by Muse are treated any differently from other plays for recommendations or reporting, and neither Meta nor OpenAI has explained how their agents choose between tracks when a request is open-ended. Whether streaming services will eventually pay or be paid for priority inside these assistants is also unknown. Those answers will decide how much of a gatekeeper the agent layer becomes.

None of this is new in kind. Voice assistants have played music on request for years. What is new is that these agents are designed to act across many services and to anticipate requests, and that the largest AI companies are competing to be the assistant a person uses all day.

What it means for independent artists

The practical advice is unglamorous: complete and accurate metadata. Correct songwriter credits, genre and mood tags, lyrics where platforms accept them, and honest AI disclosure all give agents and streaming services more to work with when matching a track to a request. Discovery is becoming conversational on the licensing side as well; Starchild lets artists describe the song they need and search a catalog of human-written songs, each carrying an AI-transparency label, with alternate versions produced through licensed style models that share revenue with the artists behind them.

Sources