ZenotiHelp

Greetings and opening message

ZTZenoti Docs Team Updated October 16, 2026 10 min readGuide

Configure how the AI voice agent introduces itself and speaks to callers β€” including greeting text, language, voice, speech speed, and pronunciation.

Audience: Center administrators who set up or maintain the AI voice agent configuration.


On this page


Before you begin

  • Your center must have an active AI voice agent. If you have not completed onboarding, see Onboarding, activation and agent status.
  • You must be authenticated in the AI Receptionist dashboard to edit agent settings.
  • Greeting changes take effect on the next call after saving. Active calls are not affected.
  • Greetings and language settings are scoped per agent. Changes to one agent's inbound configuration do not affect any outbound agent, and vice versa.

Steps

Configure inbound greetings

Inbound agents have two greeting fields β€” one for calls received during business hours and one for calls received outside business hours. The agent selects between them automatically based on the center's configured business hours at the time of the call.

  1. In the AI Receptionist dashboard, open the agent you want to configure.
  2. Select Edit AI Receptionist.
  3. Locate the Business hours greeting field. The helper text reads Message to play during business hours.
  4. Enter the text you want the agent to speak when a caller reaches the agent during your center's business hours.
  5. Locate the Non-business hours greeting field. The helper text reads Message to play outside business hours.
  6. Enter the text you want the agent to speak when a caller reaches the agent outside business hours.
  7. Select Save.

Note: If your primary language is not English, the greeting text is translated before it is spoken. If translation cannot be completed within five seconds, the agent speaks the original English text instead. Write greetings in English for the most reliable experience.

Configure outbound opening messages

Outbound agents use a different set of opening messages depending on the agent's purpose. All outbound agents have a Guest Verification Message β€” the opening statement the agent uses to confirm the guest's identity at the start of the call. Agents configured for appointment confirmations also have an Appointment Confirmation Message, and agents configured for membership dues recovery have a Membership Dues Message.

  1. In the AI Receptionist dashboard, open the outbound agent you want to configure.
  2. Select Edit AI Receptionist.
  3. Locate the Guest Verification Message field. The helper text reads The agent uses this message to verify the guest's identity at the start of an outbound call.
  4. Enter the opening statement the agent will use to introduce itself and ask the guest to confirm their identity.
  5. If the agent's Purpose is Appointment Confirmations, locate the Appointment Confirmation Message field and enter the message the agent will use when confirming an upcoming appointment with the guest.
  6. If the agent's Purpose is Membership Dues Recovery, locate the Membership Dues Message field and enter the message the agent will use when informing guests about outstanding membership dues.
  7. Select Save.

Tip: Keep the guest verification message brief and clear. The agent uses this text verbatim at the very start of the call before any guest response is received.

Add placeholder variables to greetings

Greeting fields support placeholder variables that the agent replaces with actual values before speaking. Use these to personalize the greeting with context specific to the call.

The supported placeholders are:

Placeholder Replaced with
{{agent_name}} The configured name of the AI voice agent
{{center_name}} The name of the center receiving or placing the call
{{organization_name}} The name of the organization
{{guest_name}} The guest's full name
{{first_name}} The guest's first name
{{last_name}} The guest's last name

Worked example: A business hours greeting that reads "Hello {{first_name}}, thank you for calling {{center_name}}. I'm {{agent_name}}, your AI receptionist. How can I help you today?" is spoken as "Hello Jamie, thank you for calling Luxe Spa. I'm Aria, your AI receptionist. How can I help you today?"

Set the primary language

The primary language determines the language the agent uses by default throughout the call.

  1. In the AI Receptionist dashboard, open the agent you want to configure.
  2. Select Edit AI Receptionist.
  3. Locate the Language field. The helper text reads Default language the agent will communicate in.
  4. Select the language from the dropdown. The supported languages are: English, Spanish, French, German, Hindi, Russian, Portuguese, Japanese, Italian, and Dutch.
  5. Select Save.

Note: English is the default language. Greetings are always authored in English and translated as needed before the agent speaks them.

Add additional languages

You can configure the agent to switch to additional languages during a call. When a caller speaks one of the agent's enabled languages, the agent switches to that language immediately and without comment. If the caller uses a language the agent does not support, the agent does not switch and only acknowledges the language if the caller explicitly asks.

  1. In the AI Receptionist dashboard, open the agent you want to configure.
  2. Select Edit AI Receptionist.
  3. Locate the Additional languages field. The helper text reads Specify additional languages which callers can choose from.
  4. Select one or more additional languages from the list. The same ten languages available for the primary language are available here.
  5. Select Save.

Important: Each AI voice agent handles one primary language plus any additional languages you enable. For a deployment where two completely separate phone numbers must serve different language communities, set up two separate agents β€” one per dedicated phone number.

Configure the agent voice

The agent voice controls how the AI voice agent sounds to callers. You can configure a single voice that applies to all languages, or assign different voices per language using multi-voice support.

  1. In the AI Receptionist dashboard, open the agent you want to configure.
  2. Select Edit AI Receptionist.
  3. Locate the Voice field and select the voice you want the agent to use.
  4. Select Save.

Configure per-language voices

Multi-voice support lets you assign a specific voice to each language the agent supports, so callers hear an accent that matches the language they are speaking. If no voice is assigned to a specific language, the agent falls back to the agent-level voice.

  1. In the AI Receptionist dashboard, open the agent you want to configure.
  2. Select Edit AI Receptionist.
  3. Locate the Multi-voice support section. The description reads Default voice is used for all languages. You can override this to assign a specific voice per language for a more personal touch and authentic accent.
  4. To assign a voice to a language for the first time, select Add Voice.
  5. To change an existing per-language voice assignment, select Edit Voice next to that language.
  6. Select the voice and the language it applies to, then confirm.
  7. Select Save.

Adjust speech speed

The Speed control governs how fast the agent speaks. Values below 1.0 slow the speech down; values above 1.0 speed it up. Extreme values may reduce the quality of generated speech.

  1. In the AI Receptionist dashboard, open the agent you want to configure.
  2. Select Edit AI Receptionist.
  3. Locate the Speed control.
  4. Adjust the value to your preferred pace.
  5. Select Save.

Tip: Start close to the default value and adjust incrementally. Call the agent's test phone number after each change to confirm the pace sounds natural.

Configure interrupt sensitivity

Interrupt sensitivity controls how easily a caller's speech interrupts the agent while it is speaking. A higher sensitivity means the agent stops speaking sooner when it detects the caller talking; a lower sensitivity makes the agent less likely to yield mid-sentence.

Note: Interrupt sensitivity governs how the agent responds to the caller speaking over it. It does not filter non-speech background noise such as music or ambient sound.

  1. In the AI Receptionist dashboard, open the agent you want to configure.

  2. Select Edit AI Receptionist.

  3. Locate the interrupt sensitivity setting and select one of the three values:

    • High β€” The agent yields quickly when it detects the caller speaking. Use this if callers frequently need to interrupt to correct or redirect the agent.
    • Medium β€” A balanced setting suitable for most centers.
    • Low β€” The agent is less likely to stop mid-sentence. Use this in environments with more background noise where false interruptions are a concern.
  4. Select Save.

Note: If no interrupt sensitivity is set on the agent, the platform default applies. Where no platform default is configured, the telephony provider's default β€” high β€” is used.

Write additional instructions

The Additional instructions field lets you give the agent free-text guidance that supplements its default behavior β€” for example, how to handle specific guest requests, what to say when a service is unavailable, or any custom policy the agent should follow.

  1. In the AI Receptionist dashboard, open the agent you want to configure.
  2. Select Edit AI Receptionist.
  3. Locate the Additional instructions field.
  4. Enter your instructions as plain text.
  5. To work in a larger editing area, select the expand icon. The helper text reads Click the expand icon to open fullscreen editor for detailed instructions.
  6. Select Save.

Upload pronunciation dictionaries

Pronunciation dictionaries let you define how the agent pronounces specific words β€” such as service names, provider names, or location names that the agent might otherwise mispronounce. Upload dictionaries as CSV files. A maximum of three pronunciation dictionary files can be uploaded per agent.

  1. In the AI Receptionist dashboard, open the agent you want to configure.
  2. Select Edit AI Receptionist.
  3. Locate the Pronunciation dictionaries section.
  4. Select Upload file.
  5. Select a CSV file from your computer and confirm the upload. The supported format label reads Supported formats: CSV.
  6. Repeat for additional files. The dashboard displays a warning if you attempt to upload more than three files: Maximum 3 pronunciation dictionary files allowed.
  7. Select Save.

Tips

  • Test every greeting change before going live. Use the agent's testing phone number to call in and hear exactly how the greeting sounds after saving. Adjust phrasing or punctuation if the spoken result differs from your intent.
  • Use {{first_name}} rather than {{guest_name}} for a warmer tone. First-name greetings sound more personal on a phone call, particularly for returning guests.
  • Match your business hours greeting to your in-center tone. If your front desk team is formal, keep the greeting formal. Guests notice when the AI voice agent sounds inconsistent with the rest of your brand.
  • Keep additional instructions focused and specific. Broad instructions like "be helpful" have less effect than specific ones like "if a guest asks about parking, tell them there is free parking in the adjacent lot on Main Street."
  • Upload pronunciation dictionaries before your first live call. Mispronounced service or provider names create a poor first impression. CSV entries take effect on the next call after saving.
  • Choose interrupt sensitivity based on your guest profile. Centers with technically sophisticated guests or a fast-paced front-desk style may prefer high sensitivity so callers can redirect the agent quickly. Centers with noisier environments β€” such as open-plan spas with music playing near the phone β€” may find medium or low reduces false interruptions.

Troubleshooting

The agent is speaking in English even though I set a different primary language.

The greeting is translated before the agent speaks it. If translation cannot complete within five seconds, the agent falls back to English. This can occur if the translation service is briefly unavailable. Check that your greeting text is authored in English β€” the translation happens automatically. If the issue persists across multiple calls, contact Zenoti Support.

The placeholder variable is being spoken out loud rather than replaced.

Ensure the variable name exactly matches the supported list β€” for example, {{first_name}} not {{firstname}} or {{ first_name }}. Extra spaces inside the double braces or a typo in the variable name will prevent substitution. Correct the greeting text and save again.

Callers who speak a supported additional language are not being switched to that language.

Language switching is triggered when the caller uses the language during the call. Confirm that the language is enabled in the Additional languages field and that you saved after adding it. If the agent still does not switch, test by calling and speaking several sentences clearly in the target language from the start of the call.

The uploaded pronunciation dictionary does not seem to have any effect.

Confirm the file is in CSV format and that the upload completed successfully β€” the dashboard shows a confirmation message when the upload succeeds. Pronunciation dictionary changes apply from the next call onwards, not to calls already in progress. If mispronunciation continues, verify the CSV structure matches the expected format and re-upload.

The per-language voice is not being used β€” the agent always uses the default voice.

Check that the language for which you configured a per-language voice is also enabled in the Additional languages or Language fields. A per-language voice assignment has no effect if that language is not active on the agent.

Callers are triggering the agent to stop speaking when there is only background noise.

Set interrupt sensitivity to Medium or Low. This raises the threshold at which the agent treats incoming audio as a deliberate interruption. Note that this setting affects only barge-in detection from speech β€” it does not suppress background noise from the call itself.

Was this article helpful?
Your feedback helps us improve our documentation.