Cursor has shipped Voice Chat with Agents, a two-way realtime voice call with coding agents, according to its 6 November changelog. Instead of typing prompts and reading replies, developers can now describe a change out loud, ask what the agent is doing, and hear progress updates while the work moves forward.

The feature works in the Agents Window, the IDE, on the web, and in Cursor for iOS. Starting a call is deliberately simple: when the agent is idle and the chat input is empty, a waveform button labelled "Start voice chat" appears where the Send button usually sits, next to the microphone button used for dictation. On iOS, you can also say "Hey Siri, call Cursor" to start a call with a new agent, and mute or hang up from the Lock Screen using a Live Activity.

Once the call is live, the voice agent listens, replies, and hands coding work to the agent in your chat. While that agent runs, the conversation keeps going — you can ask what changed in a diff, check on a run, search earlier messages and calls with that agent, or switch between workers in a project that has them. The coding agent also speaks for itself: it shares short progress notes while it works and a brief answer when it finishes, so you usually hear a short update instead of the full written reply.

Voice chat is powered by Grok Voice, with a choice of voices in Voice settings, including Ara, Eve, Leo, Rex and Sal. Cursor says that with Privacy Mode on, voice chat runs with zero data retention — a detail worth confirming before enabling it on regulated or client work, since voice calls are a new surface for potentially sensitive code and conversation to pass through.

It is important not to confuse this with dictation. Voice chat is separate from the older Voice Mode, which simply turned speech into text in the chat input; dictation still exists, and Cmd+Shift+Space still toggles it in the IDE. Voice chat is a two-way conversation where the agent answers out loud — closer to pair programming with a colleague than to typing with your voice.

Why it matters is straightforward: steering agents by voice changes how teams interrupt and monitor long runs without typing, including from a phone. A developer can start a long refactor from the couch, check in from the kitchen, or ask for a status update while walking — the agent keeps working and narrates as it goes. For solo developers, it removes the friction of switching between keyboard and explanation; for teams, it turns agent supervision into something you can do away from the desk.

There are limits to keep in mind. Voice is a new interaction surface, not a model upgrade: the underlying coding agent is the same, and spoken instructions can be vaguer than written ones, so precise specs may still be better typed. And the Privacy Mode claim — zero data retention — is Cursor's own statement; teams with compliance requirements should verify the policy before rolling it out. But for anyone already living in Cursor, the update is a one-click try: update Cursor and the waveform button appears.

The timing is notable: voice is arriving in coding tools from several directions at once, and Cursor's bet is that the agent itself should do the talking. Short spoken progress notes replace the need to watch a terminal or scroll a chat log, and the "Hey Siri, call Cursor" hook turns a phone into a remote control for a running agent. The trade-off is precision — spoken instructions are naturally looser than written ones — so the feature is likely to shine for monitoring, quick redirects and status checks, while detailed specs still belong in text. For teams already on Cursor, it is a free upgrade away: update the app and the waveform button is there.

Sources