Skip to main content
Talk mode lets you call your agent and speak to it instead of typing. The agent answers out loud, and when you ask for something that needs real work, such as looking up a ticket, sending a message, or running a skill, it does that work in the chat while you stay on the call. Everything said on the call is written into the chat as ordinary messages, so a call is just another way to use a task you can scroll back through later.

​
Turn on voice calls

Gumball, your personal agent, can always be called. Custom agents have to opt in, which the agent’s editors do from the Settings tab, under Voice.
Voice section of an agent's settings with Allow voice calls turned on and the Voice model dropdown open, listing GPT Live and Grok Voice 3
Voice minutes use credits, and the rate depends on the voice model. See Voice call costs for the rates and how they appear in a chat’s breakdown.

​
Start a call

Open a chat with the agent and select the voice button at the right of the message bar, next to the microphone.
Message bar with the plus menu, the model name, the microphone button for dictation, and at the far right the round voice button that starts a call
The two buttons do different things. The microphone dictates a single message: your audio is transcribed and sent as text. The voice button starts a live call where the agent talks back.
You can start a call in a brand-new chat or in one you have already been typing in; the agent hears the recent conversation either way.

​
While you are on a call

Message bar during a live call: the message box stays available above a row with the plus menu, the agent's avatar for voice settings, a live waveform, your avatar, speaker and microphone mute buttons, and an End call button
  • Your words appear in the chat. Each thing you say is added as a message tagged Voice call, and the agent’s spoken replies are added as its messages.
  • Small talk gets answered on the call. Real requests go to the agent, which then runs in the chat exactly as if you had typed them, with thinking, tool calls, and steps all visible.
  • You can keep talking while work runs in the background. Longer requests run as background work, so a new question doesn’t cancel them. When the work finishes, the agent can speak its result during the call, or post it in the chat as text if you’ve hung up.
  • The agent can look back at earlier calls. Ask something like “what did we decide on our last call?” and it can search your earlier conversations with it.
  • The agent can hang up. When the conversation is done, for example after you say goodbye, the agent can end the call once it has finished speaking.
  • The agent tells you what it can do. It knows its connected apps, skills, and knowledge, so you can ask “can you check Linear?” and get a truthful answer.
  • The live call appears in the message bar, with a live waveform and controls to Mute speaker, Mute microphone, and End call. Select the agent’s avatar for Voice settings.
  • You can type and attach files during the call. Type in the message box, or use the + menu to add files. The agent handles typed messages and attachments like spoken requests, in call order.
Ending the call leaves the chat exactly where it was: the transcript, the agent’s work, and any files it produced stay in the task.

​
Good to know

  • One call per task. A chat can hold one live call at a time. If several calls are already running on your account, the next one is refused until you end one.
  • Model access applies. If your organization restricts which models you can use, the voice model may be blocked for you. The call says so, and you can request access from an admin.
  • Credits run the call. You need credits for the minute ahead, so a call ends if your balance runs out mid-conversation.
  • Calls happen in the browser. You start a call from the agent’s chat in Gumloop.
  • Silence ends a call. After about two minutes of silence with nothing running, the call ends. It stays open while background work for the call is still running.
  • Incognito calls can’t run background work. In an incognito task, the agent finishes each request before it answers.

​
Voice memos

An agent can reply with a voice memo: a short spoken audio message, delivered as an MP3 file. Ask for one (“send me that as a voice memo”), and the agent writes the script and Gumloop reads it aloud in the agent’s Talk mode voice. If that voice isn’t available for memos, Gumloop uses Marin. Voice memos use credits for the speech and follow your organization’s model access. If Gumloop can’t make the memo, the agent sends its reply as text instead.

​
FAQ

No. Voice follows chat access: if someone can chat with the agent, they can call it once Allow voice calls is on.
Yes. The call is transcribed into the chat as it happens, so the task holds both sides of the conversation along with everything the agent did.
By time, not by tokens, and it shows up as Voice in the chat’s credit breakdown. Work the agent does during the call (tool calls, thinking, compute) is billed the usual way on top. See Voice call costs.
The microphone records one message and sends it as text. The voice button opens a live, two-way call.