Setting Up Voice AI: Your 24/7 Virtual Receptionist

Guides & Tips
, , , ,
1

Exoserva Voice AI lets you deploy a virtual receptionist that answers calls, books appointments, and handles customer inquiries around the clock. Whether your team is on a job site or it is 2 AM, your business never misses a call. This guide walks you through provisioning a phone number, creating an AI assistant, selecting a voice, and fine-tuning its behavior so it sounds and acts exactly the way you want.

:clock1: Estimated time: 15-20 minutes

Before You Begin

  • An active Exoserva account with admin or owner permissions
  • Your business hours and routing preferences decided (which calls go where)
  • A general idea of the tone and personality you want your AI assistant to have

Step 1: Navigate to Voice AI Settings

Open the left sidebar and go to Settings. Click the Phone Numbers tab (you may also see it labeled Voice depending on your plan). You will land on the Voice AI settings page with multiple configuration tabs across the top: Status, Assistants, Voice, Behavior, Business, Tools, Context, Analytics, Phones, and Calls.

Step 1: Navigate to Voice AI Settings
Step 1: Navigate to Voice AI Settings1280×800 128 KB

:bulb: Tip: If you do not see the Phone Numbers or Voice tab, your plan may not include Voice AI. Contact support@exoserva.com to enable it.

Step 2: Provision a Phone Number

Click the Phones tab. Here you can provision a new phone number for your Voice AI. Click Add Phone Number and choose your area code or region. Exoserva will assign a local or toll-free number that customers can call. Once provisioned, the number appears in your phone number list and is ready to be linked to an assistant.

:bulb: Tip: Choose a local area code that matches your service area. Customers are more likely to answer and call back local numbers.

:warning: Warning: Phone number provisioning may take a few moments. Do not refresh the page while the number is being set up.

Step 3: Create a Voice Assistant

Switch to the Assistants tab and click Create Assistant. Give your assistant a name (for example, “Sarah” or “Alex”) and write a short personality description. The personality field tells the AI how to introduce itself and what demeanor to adopt. For example: “Friendly and professional HVAC receptionist who speaks clearly and keeps conversations focused on scheduling and service inquiries.” You can create multiple assistants for different purposes, such as one for after-hours and another for sales inquiries.

:bulb: Tip: Pick a name that feels natural for a phone conversation. Short, common names work best because they are easy for callers to remember and the voice engine pronounces them clearly.

Step 4: Select a Voice Provider and Model

Go to the Voice tab to choose how your assistant sounds. Exoserva supports several voice providers: ElevenLabs, Azure, PlayHT, Deepgram, and Cartesia. Each provider offers different voice models with varying characteristics – some sound warmer, others more crisp and corporate. Select a provider, then browse or preview the available voices. You can also select the language your assistant speaks, which is especially useful if you serve bilingual communities.

:bulb: Tip: ElevenLabs voices tend to sound the most natural for conversational phone calls. If latency matters most (fast response time), Deepgram is an excellent choice. Try a few and listen to the previews before deciding.

Step 5: Configure Behavior Settings

Click the Behavior tab. Here you control the personality traits of your assistant in more detail. Adjust the tone (formal, friendly, casual), verbosity (concise responses vs. detailed explanations), and language style. You can also set how the assistant handles interruptions, how long it waits before responding, and whether it confirms details back to the caller. These settings let you match the assistant to your brand voice.

:bulb: Tip: For most field service businesses, a friendly but professional tone with concise responses works best. Customers calling about a broken AC or a plumbing emergency want quick answers, not long explanations.

Step 6: Set Business Rules and Routing

Navigate to the Business tab. This is where you define how the assistant handles calls based on your business logic. Set your business hours so the assistant knows when to offer live transfers vs. taking messages. Configure call routing rules – for example, urgent calls can be forwarded to an on-call technician while routine scheduling requests are handled entirely by the AI. You can also define escalation paths for situations the AI cannot resolve.

:warning: Warning: Make sure your business hours are accurate. If you set incorrect hours, the assistant may try to transfer calls to your team when nobody is available, leading to a poor customer experience.

Step 7: Add Knowledge Context

Open the Context tab. This is where you give your assistant the knowledge it needs to answer customer questions accurately. Add information about your services, pricing ranges, service area, common FAQs, and any special policies (such as cancellation terms or warranty information). The more context you provide, the better the assistant can handle real conversations without needing to transfer to a human. You can paste text, add structured Q&A pairs, or upload reference documents.

:bulb: Tip: Start with the five most common questions your receptionist or office staff answers every day. Things like “What areas do you serve?”, “How much does a service call cost?”, and “Do you offer emergency service?” are great starting points. You can always add more context over time.

Step 8: Make a Test Call

Before going live, test your assistant. Call the provisioned phone number from your personal phone and have a realistic conversation. Try asking about your services, requesting an appointment, and asking a question that is not in your knowledge base (to see how the assistant handles unknowns). Listen for natural pacing, correct information, and appropriate escalation behavior. Go back to any settings tab and adjust if something does not sound right.

:bulb: Tip: Have a team member who was not involved in the setup make a test call as well. Fresh ears catch things you might miss, like awkward phrasing or missing information.

Step 9: Review Call Analytics

After your assistant has handled a few calls (or even just your test calls), check the Analytics tab for performance data and the Calls tab for individual call logs. Analytics shows metrics like total calls handled, average call duration, successful bookings, and transfer rates. The Calls tab lets you review individual call recordings and transcripts. Use these insights to refine your assistant settings, update its knowledge base, and improve the caller experience over time.

:bulb: Tip: Review call logs weekly for the first month. Look for patterns in questions the assistant could not answer and add that information to the Context tab. Most businesses see significant improvement in AI performance after two to three rounds of refinement.

What’s Next?

Now that you’ve completed this guide, check out:

Need help? Post in the Tech Support category or contact support@exoserva.com.