Sonos Adds Presets and Early Access to Its 27voice AI Assistant
Sonos is rolling out an app update that adds Presets and early access to its AI voice assistant, 27voice. The Verge reported the update on November 3, 2026.
Presets are saved playback configurations. Each one plays music at a set volume in a specific room, and each appears on the app homescreen for one-tap playback. Sonos Support documents the feature.
US customers can sign up for the 27voice waitlist on November 6, with access starting November 12. The Early Access version supports US English, and customers in the United States will opt in to try it. Sonos had previously said Early Access would begin in fall 2026.
According to Sonos, 27voice can start music, move playback to another room, group and ungroup speakers, and set timers. The company also says the assistant handles complex natural-language requests (phrased the way people actually speak, rather than as exact commands), such as playing an album described by its cover. Those requests depend on how well the assistant maps a loose description onto a music service's catalog, which is where voice control on connected audio has historically struggled.
The assistant sits on top of Sonos 27, the software update the company announced on September 8, 2026. Sonos said the release provides AI control through outside assistants and agents (software that carries out multi-step tasks on a user's behalf). The company describes Sonos 27 as software that turns every speaker into one system, linking speakers, services, and rooms. Sonos 27 also lets portable speakers automatically become home theater surrounds, according to the investor announcement. The Verge reports that Sonos plans to let users create up to 10 custom agents.
The current app update includes several smaller changes. Users can swipe on the Now Playing screen or the mini player to skip tracks, and they can rearrange their favorites. The app also works with System Health, an iOS-only feature that helps identify problems in a speaker setup.
The broader context here is a shift in where control lives. A multi-room system is, in practical terms, a graph of devices, zones, and streaming sessions that all reference one another. Voice has been a poor tool for editing that graph by hand, which is why grouping, room-to-room handoff, and per-room presets have mostly been menu tasks. Routing those actions through a natural-language layer, and opening that layer to third-party assistants and agents, moves the interface problem from the app screen to the intent parser (the component that interprets what a request means). How accurately that parser resolves room names, device states, and ambiguous album descriptions will decide whether the feature saves time or adds a step. Early Access, with its opt-in model and waitlist, suggests Sonos expects that behavior to change as it gathers feedback.
In my view, the more practical question for owners is less about the chatbot than about the plumbing. Presets give a deterministic path for routines people repeat every day, such as the same playlist, room, and volume each morning. A voice layer that can reach those presets, group speakers on demand, and set timers without a screen is a useful addition to a home audio system, provided it fails gracefully when it misunderstands a request. The real tests are the messy cases: a half-remembered album cover, two rooms with similar names, and a household where more than one person is issuing commands.
What the update enables is a more direct route from intent to sound. A user who wants the kitchen playing a specific record at a comfortable level can ask for that without opening an app, and a user who wants the whole house to follow one track can do it in a single request. Whether 27voice reaches that standard for most households will become clearer once the November 12 access date arrives and early users start reporting results.
