Channel settings attributes

These Level C attributes control channel-level behavior across chat, voice, and SMS — including how messages are formatted, how conversations are routed by source, and which provider handles outbound SMS. They are hidden from the standard Attributes page and only appear when you enable the Show Hidden toggle.

Conversation tools

IDN: project_attributes_settings_newo_tools

This attribute defines the tools available to the AI Employee during a conversation. You can use it to override existing built-in tools or introduce entirely new ones — for example, a custom webhook that fires when a user provides their phone number, or a modified calculation_tool with different trigger conditions.

When this attribute is empty, the AI Employee uses built-in tools only. When you provide a tools object, the runtime merges your custom definitions with the built-in set. If a custom tool key matches a built-in key, your definition takes precedence.

Each tool requires a conditions object (which determines when the tool fires) and a useConversationHistory boolean. The action field is optional; if omitted, the runtime defaults to send_event.

Supported action types:

• send_event: Fires a named system event. No after-actions are supported.
• send_webhook: Makes an HTTP request to an external endpoint. Supports chained after-actions (send_urgent_message, set_persona_attribute, set_prompt_section) that run in order when the webhook returns a non-empty response body.

Conflict resolution: Use conflictPolicy.tag to group mutually exclusive tools. If more than one tool in the same tag group is selected for execution, the runtime reranks them and executes only one.

::: ⚠️ CAUTION

This feature is under active development. Configuration options and behavior may change in future releases. Validate each tool individually before adding more, and confirm trigger behavior in a test environment before deploying. :::

Conversation sources map

IDN: project_attributes_setting_conversation_sources_map

This attribute maps connector IDs to short, stable source labels that the AI Employee can use to adapt its behavior based on where a conversation originated. Once a mapping is defined, conversation.source becomes available as a meta value in your prompt logic.

For example, you could instruct the AI Employee to ask for a phone number when the source is website_chat, but tell the user a manager will follow up via Messenger when the source is facebook.

Example mapping:

{
  "manychat_facebook_connector": "facebook",
  "manychat_tiktok_connector": "tiktok",
  "website_chat": "website"
}

Use short, stable labels as values and match the keys to the connector IDs shown on your Integrations page. The resolved source is cached in the persona state for the duration of the session and is also included in end-session data for downstream reporting and integrations.

When the map is empty or no matching connector is found, conversation.source falls back to the conversation channel label.

Chat — split long messages

IDN: project_attributes_setting_chat_split_messages

When enabled, the AI Employee splits certain chat replies into two messages: the main explanation is sent first, followed by any trailing question as a separate message. This applies only to chat-channel flows and only when a reply ends with a question and contains multiple sentences — otherwise the AI Employee sends a single message regardless of this setting.

The split path adds a short delay between message parts for natural pacing. Voice and webcall channels are not affected by this setting.

Values:

• True: The AI Employee uses the split-message path for qualifying chat replies.
• False: All chat replies are sent as a single message.

Check spam or voicemail

IDN: project_attributes_setting_check_spam_or_voicemail

When enabled, the AI Employee actively classifies ongoing voice conversations to detect spam calls and voicemail systems. On a positive detection, the runtime marks the relevant persona state and can issue a stop-call command.

Classification can return True, False, or Skip (for example when the system detects a screening service or a busy announcement rather than a voicemail). For outbound sessions, voicemail detection can run even when this attribute is set to False.

Values:

• True: Spam and voicemail classification runs on voice transcription events, and automated handling actions can be triggered.
• False: Inbound spam checks are skipped. Outbound voicemail detection may still run independently.

::: 🗒️ NOTE

The stop-call command path used after a positive detection depends on project_attributes_setting_voice_integration_service. Ensure that attribute is configured correctly before enabling spam or voicemail checks. :::

Phone — dialpad output

IDN: project_attributes_setting_phone_dialpad_output_enabled

When enabled, the AI Employee can send DTMF (dialpad) digits during a phone conversation — for example, to navigate an IVR menu during an outbound call flow. When the conversation context indicates dialpad output is needed, the runtime sends a dial_digits command to the voice connector.

If the channel or provider prerequisites are not met, the runtime exits without sending dialpad output even when this attribute is enabled. Only enable this for phone workflows that have been explicitly tested for keypad interaction.

Values:

• True: The AI Employee can attempt dialpad output when the conversation context requires it.
• False: The dialpad output path is blocked.

::: ⚠️ CAUTION

Dialpad output is only available for supported voice integration providers. Verify compatibility with your project_attributes_setting_voice_integration_service setting before enabling. :::

SMS provider

IDN: project_attributes_setting_sms_provider

This attribute selects the backend delivery path used when the AI Employee sends outbound SMS messages. It is evaluated by the SMS worker for both command-send and actor/message-send flows.

Possible values:

• twilio_messenger: Routes SMS through the Twilio Messenger connector path.
• sms_service: Routes SMS through the dedicated SMS service connector path.

Keep this value aligned with your active SMS connector setup. Mismatches between this attribute and your configured connector will cause SMS delivery to fail.