> ## Documentation Index
> Fetch the complete documentation index at: https://docs.dovetail.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Voice mode in Chat

> Talk to Chat or a Digital Twin out loud and share your screen for context. Voice mode is in beta and needs a workspace admin to turn it on.

<Info>
  Available on the [Enterprise plan](https://dovetail.com/pricing/)

  Voice mode is a [beta feature](/help/beta-features). A workspace admin needs to turn it on first.
</Info>

With voice mode, you ask Chat questions out loud and hear its answers spoken back. It uses the same workspace data and the same [permissions](/help/chat#who-can-see-what) as typed Chat.

Voice mode also works with [Digital Twins](/help/agents/digital-twins). You can talk to a twin the way you would talk to the customer it represents, for example to rehearse a sales pitch or test how a customer segment might react to a feature idea. You can share your screen during a voice chat, so a twin can react to a design or prototype it can see.

***

## Before you start

Voice mode is available when both of these are true:

1. **The beta is turned on for your workspace.** Workspace admins turn it on in [Settings → Beta](https://dovetail.com/settings/beta). See [Beta settings](/help/beta-features).
2. **AI processing location is set to Best available.** The model behind voice mode doesn’t yet support processing pinned to a single region. If your workspace uses United States, Europe, or Australia, voice mode isn’t available. Admins can change this in **Settings → Security → AI settings**. See [AI settings](/help/security-settings#ai-settings).

<Warning>
  Changing AI processing location to Best available applies to all AI features in the workspace, not only voice mode. For Europe and Australia, Best available may process data in the United States. Check your data residency requirements before changing it.
</Warning>

On an Enterprise trial, AI processing location is set to Best available and can’t be changed, so voice mode is available as soon as the beta is turned on.

***

## Start a voice chat

1. Open [Chat](/help/chat).
2. Select the voice mode button, marked with a waveform icon, in the prompt box.
3. If your browser asks for microphone access, allow it.
4. Start speaking. Chat replies out loud.

Chat keeps the context of wherever you started, just as it does for typed questions. A voice chat started inside a project answers about that project.

***

## Talk to a Digital Twin

1. Open Chat and click the Digital Twin icon to enter the twin’s persona. See [Talking to your twin](/help/agents/digital-twins#talking-to-your-twin).
2. Start voice mode.
3. Speak to the twin as you would to the customer. It answers in character, drawing on the data attached to it.

A twin can only speak to what is in its data. If no customer has discussed a topic, the twin should say so rather than invent an answer. See [Digital Twins](/help/agents/digital-twins).

***

## Share your screen

During a voice chat, you can share your screen so Chat or a Digital Twin can see what you are looking at. This is useful when the question is about something visual, like a design, a prototype, a dashboard, or a document you haven’t imported into Dovetail.

1. Start a voice chat.
2. Select the screen share button in the voice chat controls.
3. In your browser’s sharing dialog, choose a whole screen, a single window, or a browser tab. Depending on your browser, you can also share the audio that’s playing.
4. Keep talking. You can refer to what’s on screen, for example *“What would you change on this page?”*

To stop sharing, use the screen share button again or your browser’s **Stop sharing** control.

Screen sharing isn’t available on mobile devices, because mobile browsers don’t support it. Voice chat without screen sharing still works on mobile.

<Warning>
  Everything visible in the area you share is sent to the AI model along with your voice, including notifications, other tabs, and personal information. Share a single window or tab rather than your whole screen when you can, and close anything you don’t want included.
</Warning>

### Use screen sharing with a Digital Twin

Share your screen with a twin to get a customer segment’s reaction to something they can see, rather than something you describe.

* **Run a usability test.** Share a prototype and ask the twin to talk through what it would do first, what it expects to happen, and where it gets stuck.
* **Review a design.** Share a mockup and ask what lands, what’s confusing, and what the twin would push back on.
* **Check a page or message.** Share a landing page, pricing page, or email draft and ask whether the language matches how the segment describes its problems.

The twin’s reactions are grounded in what customers in that segment have already said. They show you where to look before you test with real customers, not a replacement for doing so.

***

## Use cases

Voice mode suits work that happens out loud: rehearsing a conversation, thinking through a problem, or getting briefed while your hands are busy. Most of the use cases below pair voice mode with a Digital Twin. For twin instructions you can copy, see [Digital Twin examples](/help/agents-use-cases#digital-twins).

### Rehearse a sales or renewal call

Build a twin from a single account’s call history, or from a segment such as enterprise buyers in one industry. Then run the conversation you are preparing for. Sales calls contain the objections that account or segment has already raised, so the twin can raise them too.

* *“We’re looking at a competitor that’s cheaper. Why should we stay?”*
* *“I need to take this to procurement. What will they ask me?”*

This works best when your calls are connected to Dovetail and enriched with CRM data. See [Revenue intelligence](/help/revenue-intelligence).

### Practice objection handling

Ask the twin to push back rather than agree. Instructions like *“push back when something doesn’t land — don’t hedge to be helpful”* produce a more realistic practice session. New reps can use the same twin to hear how a segment describes its problems before their first live call.

### Prepare for a churn or save conversation

A churned customer twin draws on the calls, tickets, and highlights recorded before the account left. Customer success teams can talk through what broke, what the customer tried to signal, and what might have kept them, then apply that to an at-risk account.

### Test a concept or message

Describe a feature, a pricing change, or a positioning statement to a persona twin, or share it on screen, and ask how it lands. Twins are useful for language questions — *“What would you call this?”* — and for finding the first objection a segment would raise.

### Pilot an interview guide

Before a research session, ask your interview questions out loud to a twin of the segment you are about to interview. If the twin already has a well-evidenced answer, you can spend the live interview on something the data doesn’t cover.

### Brief yourself before a meeting

Voice mode works without a twin, too. Ask Chat for a spoken summary on your way into a meeting:

* *“What has this account raised in the last three months?”*
* *“What are the top three things churned customers mention?”*

<Tip>
  A twin summarizes customer data, so its words may not match any customer’s exact words. Before you quote a twin in a deck or a decision, open the source it drew from. See [Digital Twins](/help/agents/digital-twins).
</Tip>

***

## Troubleshooting

<AccordionGroup>
  <Accordion title="I don’t see the voice mode button">
    Check that a workspace admin has turned on the voice mode beta and that AI processing location is set to Best available. Voice mode isn’t available on Free plans, because beta features aren’t available on Free plans.
  </Accordion>

  <Accordion title="I can’t share my screen">
    Screen sharing isn’t available on mobile devices. On a computer, check that your browser has permission to record your screen. On macOS, this is set in **System Settings → Privacy & Security → Screen & System Audio Recording**, and you may need to restart your browser after allowing it.
  </Accordion>

  <Accordion title="Chat can’t hear me">
    Check that your browser has permission to use your microphone, that the right microphone is selected in your system settings, and that no other application is using it.
  </Accordion>
</AccordionGroup>

***

## FAQs

<AccordionGroup>
  <Accordion title="Can I use voice mode if my workspace requires a specific AI processing region?">
    Not yet. Voice mode requires AI processing location to be set to Best available.
  </Accordion>

  <Accordion title="Does voice mode see anything typed Chat can’t?">
    No. Voice mode searches and answers only from content your account can already access.
  </Accordion>

  <Accordion title="How do I send feedback?">
    Use the thumbs-up or thumbs-down icons in the product. See [Beta settings](/help/beta-features).
  </Accordion>
</AccordionGroup>

***

## Related

<CardGroup cols={2}>
  <Card title="Chat" icon="message-circle" href="/help/chat">
    Ask questions about your customer data and get cited answers.
  </Card>

  <Card title="Digital Twins" icon="users" href="/help/agents/digital-twins">
    Build a twin from real customer data and talk to it in Chat.
  </Card>

  <Card title="Revenue intelligence" icon="trending-up" href="/help/revenue-intelligence">
    Analyze sales and customer success calls alongside CRM data.
  </Card>

  <Card title="Agent use cases" icon="bot" href="/help/agents-use-cases">
    Example instructions for agents and Digital Twins.
  </Card>
</CardGroup>


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.