Skip to main content

Configuring Your AI Agent’s Voice, Personality, and Greeting

Learn how to fine-tune how your AI Agent sounds and interacts with callers.

Emily avatar
Written by Emily
Updated today

Once your AI Agent is created, you can customize how it speaks, what it says first, and how it interacts with your callers.

Changing the Agent’s Voice and Language

  1. Go to AI Agents in your Textedly account.

  2. Click on the AI Agent you want to edit.

  3. Go to Voice

  4. Choose:

    • Persona: Male or female

    • Voice: Select from the available voices (gender, tone, accent).

Click Save to apply the new voice. Changes usually take effect on the next call.

Writing a Great Greeting

Your greeting is the first thing callers hear. It should be:

  • Short and clear

  • Welcoming and on-brand

  • Explicit about being a virtual assistant and clearly state “you are on a recorded line”

Controlling How Long the Agent Talks

If the Agent gives answers that feel too long or too short:

  • Add instructions like:

    • “Keep answers short—2–3 sentences max unless the caller asks for more detail.”

    • “Summarize your answer at the end with one short sentence.”

This helps ensure callers don’t feel overwhelmed or confused.

Handling Sensitive Topics

To keep your Agent safe and on-brand, clearly tell it what to avoid:

  • “Do not give legal, financial, or medical advice.”

  • “If callers ask for something you shouldn’t answer, apologize and offer to route them to a human or take a message."

  • “Never share internal-only information, passwords, or login details.”

Update your instructions (PDF files) as needed if you notice any responses you’d like to refine.

Testing and Iterating

After any change to voice, greeting, or behavior:

  1. Place a test call and try a few different scenarios.

  2. Listen carefully to:

    • How the Agent introduces your business

    • Whether the tone matches your brand

    • Whether responses are clear and correct

You can always come back and tweak the greeting or instructions as your use cases evolve.

Did this answer your question?