Skip to main content

Voice mode

The chat input has two voice controls. They look related, but they do different jobs:

  • Speech to Text (microphone) turns what you say into text in the compose box. You edit and send it like a typed message.
  • Full Voice Mode (chat bubble) is a live spoken conversation. The agent replies out loud.

You see both controls when Full Voice Mode is off. Starting Full Voice Mode hides Speech to Text.

Speech to Text

Use the microphone when you want to dictate a message, not hold a live conversation.

  1. Click the Speech to Text microphone in the chat input.
  2. Speak. A Listening... bar appears with a live preview of your words, and the mic pulses red.
  3. Click the microphone again to stop.
  4. The transcribed text is added to the compose box. Edit it if you need to, then send it as a normal message.

If there is already text in the box, the new words are appended. The transcript is capitalized and ends with a period.

If transcription takes a moment after you stop, the button shows a spinner until the text is ready.

Speech to Text is available on every plan. It is hidden while Full Voice Mode is on.

Starting Full Voice Mode

  1. Click the Full Voice Mode chat-bubble icon in the chat input.
  2. If you are chatting with a specific agent, the session starts with that agent.
  3. Start talking. The agent listens and replies out loud.
  4. Stop with the same voice controls when you are finished.

Full Voice Mode is available to admins and on Pro and Enterprise plans. If it is not on your plan, the control takes you to Upgrade.

Full Voice Mode

Full Voice Mode keeps the microphone open for a continuous back-and-forth conversation, rather than requiring you to start each turn.

Two things to know:

  • Your screen stays awake while a voice session is live, so a long conversation is not cut short by the display sleeping. The screen is released as soon as the session ends.
  • It is scoped to one browser tab. Enabling it in one tab does not turn it on in others, so a second Ordify tab will not start listening unexpectedly.

Voice and agents

A voice conversation is a normal conversation. The agent uses the same instructions, tools, knowledge, and jobs it would in text, and the transcript is saved to the session so you can return to it later.

What's next