Voice, Channels, and Integrations

Documentation Index

Fetch the complete documentation index at: https://docs.utari.ai/llms.txt

Use this file to discover all available pages before exploring further.

This guide shows where voice setup, external channels, and approved app credentials live inside your Utari instance. Voice and channels determine how members can interact with the AI outside the normal chat experience. Integrations and app credentials determine which approved connected accounts workers or workflows can use. By the end, you should know what has to be prepared before voice can go live, where channel setup lives, and how to keep connected-account details private during setup.

​

How to use this guide

Use this guide to decide what is ready to turn on and what still needs owner approval, Utari support, or a connected account before members use it. A feature can appear in the dashboard before it is fully live for members, so review the setup requirements before enabling anything.

​

What this guide covers

• Where voice setup lives.
• Why voice requires an uploaded and activated voice sample first.
• What the voice-call experience looks like once voice is active.
• Where Email, Telegram, Slack, and other channel setup lives.
• Where approved app credentials are managed.
• What connected-account details should stay private during setup.

​

Before you begin

Voice and channels can affect how people interact with your instance outside the dashboard. Integrations can affect which connected accounts workers or workflows can use. Only connect or create something when the owner and setup team know exactly what account, token, or workflow is being used. Treat bot tokens, app credentials, and access tokens like passwords. Keep them private, and only share them with the approved person handling that connection.

​

Setup map

Open the profile or workspace menu at the bottom-left of the dashboard. For this SOP, the important areas are Instance Config > Voice, Voice Calls, Integrations, and Channels.

• Use Instance Config > Voice to configure voice.
• Use Voice Calls to review or use the call experience once voice is active.
• Use Integrations / App Credentials for approved connected accounts that workers or workflows need permission to use.
• Use Channels for member-facing communication paths like Email, Telegram, or Slack.

​

Voice setup

Open Instance Config, then choose the Voice tab. Use this screen to configure voice before expecting the Voice Calls page or member-facing voice options to work. Voice is not available as a member-facing voice chat or call option until a voice sample has been uploaded and activated. If no voice sample is active, the dashboard can still show the voice setup fields, but the actual voice-call experience is not ready to use. Follow this setup order:

1. Upload the voice sample.
2. Wait for the voice clone to be available and activated.
3. Choose the Voice Agent that should handle voice calls.
4. Review the Greeting Message and Thinking Message.
5. Save the configuration.
6. Enable voice calls when the control becomes available.
7. Test the voice-call experience before sending members to use it.

On the Voice screen, review these areas:

• Voice Clones: where the owner uploads a voice sample.
• Upload Sample: accepts a 30-second to 5-minute audio sample in WAV, MP3, or M4A. Use natural conversational audio with one speaker, a quiet room, consistent input level, and minimal background music, background noise, or cross-talk.
• Voice Agent: selects which agent will be used during voice calls. The selected agent controls the worker behavior, personality, and knowledge the voice call uses.

Audio voice samples are for voice calls. This is different from SOP 3’s Voice / Style Source folders, which are writing and tone reference materials inside the knowledge layer.

​

Voice messages and enabling voice calls

Scroll lower on the Voice tab to review the voice messages and enablement controls. Use Greeting Message for the AI’s opening line when a voice call starts. Use Thinking Message for the filler message played while the AI is processing during a voice call. The Enable Voice Calls control should stay unavailable until a voice clone has been uploaded and activated. When voice is not ready yet, the helper text may say Upload and activate a voice clone first.

​

Voice Calls page

Open the profile or workspace menu, then choose Voice Calls. Review this page after the Voice tab setup has a sample uploaded, a voice clone activated, and a Voice Agent selected. If the page says voice calls are not set up, finish the voice sample and activation flow before testing voice calls or sending members to use voice.

​

After the voice sample is active

Once a voice sample is uploaded and activated, the dashboard can show a New Voice Call option and a voice-call screen for the selected worker. The worker name, avatar, and available minutes may differ by instance. If the instance runs out of voice minutes, email hey@utari.ai to refill.

​

Channels

Open the profile or workspace menu, then choose Channels. Channels are the external places people can interact with workers. Use Channels for member-facing communication paths, not for general business app permissions. The Channels page shows:

• Email: email channel option for the instance.
• Telegram: requires a Telegram bot token from BotFather.
• Slack: has an Add to Slack button for installing the Utari bot.

For Telegram, use the BotFather token field only when you have the approved bot token ready. For Slack, use the setup button only when Slack is part of the approved setup plan. Do not assume a channel is live just because it appears on the Channels page. A visible channel means it is available for setup; it still needs to be connected and tested.

​

Integrations and app credentials

Open the profile or workspace menu, then choose Integrations. This opens App Credentials. Use App Credentials when a connected business app needs an approved account connection. Start by clicking Connect New App, even if there are no existing connections yet. Connect the app here first. After the app is connected, attach or use that connection with the relevant worker or workflow as needed. If a connection opens a provider-specific credential form, review the fields carefully before saving. A Meta Ads connection, for example, may ask for:

• Connection name
• Access token
• Default ad account ID
• Set as default Meta Ads connection
• Verify
• Save connection

Do not enter or share real tokens or credentials unless you are actively completing the approved setup. If Meta Ads setup is required, connect the approved Meta account for the ad account that should power this instance. The connection is account-based: if the instance owner connects Meta Ads, Utari can only use the ad account access available through that owner’s connected Meta account. It does not grant access to unrelated accounts.

​

Decide who is allowed to connect external systems

Before voice, channels, or app credentials are used broadly, decide who owns the setup and who is allowed to manage connected accounts, tokens, credentials, and external channels. Assign clear owners for:

• Voice sample recording and upload.
• Voice agent selection.
• Voice greeting and thinking messages.
• Email, Telegram, or Slack setup.
• App credentials such as Meta Ads.
• Credential rotation and removal if access changes.

The owner should understand what each connection does, who owns it, and who can rotate or remove it later if account access changes. Utari or the approved technical owner should help when setup involves tokens, bot credentials, app credentials, or external accounts.

​

What to avoid

Do not share real bot tokens, app credentials, access tokens, or connected-account values with anyone who does not need them for the approved setup. Do not assume a channel is live just because it appears on the Channels page. A visible channel means it is available for setup; it still needs to be connected and tested. Do not expect voice chat or voice calls to appear until a voice sample has been uploaded and activated. Do not add app credentials unless there is a real setup plan and someone owns the credential afterward.

​

Quick check

You are ready to move on when you can answer:

• What is the order for making voice ready?
• What makes a usable voice sample?
• What does the selected Voice Agent control?
• Where do you review whether voice calls are set up?
• What appears once the voice sample is active?
• What should you do if voice minutes run out?
• Where do you connect Email, Telegram, or Slack?
• When should you use Channels instead of App Credentials?
• Where do you manage app credentials?
• Who on the team is allowed to manage, rotate, or remove tokens and credentials?

​

Next guide

Continue to Access, Members, Usage, Notifications, and Team Handoff when you are ready to configure member access, invites, analytics review, owner notifications, and the launch operating rhythm.