How I Built an End-to-End Voice AI Receptionist & Database Sync System in Make.com

Knowledge Hub
, , , , ,
1

Dental clinics and service-based businesses lose thousands of dollars every month to missed phone calls and manual appointment scheduling. To solve this, I built an automated Voice AI Receptionist ecosystem using Make.com, dynamic Webhooks, Airtable, Gmail, and external Voice AI APIs.

Here is a full breakdown of the architecture, key routing logic, and how to handle live data synchronization seamlessly.**

System Architecture Overview**

The system is split into two core automation flows:

  1. Inbound Booking & Status Handler: Receives live updates from the Voice AI agent and routes data dynamically based on customer intent (Booked, Rescheduled, Cancelled).

  2. Outbound Call Trigger & Status Sync: Initiates outbound voice calls directly from database records and syncs call outcomes back into the database.

Flow 1: Inbound Voice Receptionist & Dynamic Routing

Screenshot 2026-06-21 213242
Screenshot 2026-06-21 2132421366×643 95.8 KB

When a patient interacts with the AI voice receptionist, the payload is sent via custom Webhooks into Make to execute the business logic.

  • Webhook Trigger & Response: Captures incoming voice agent payloads and returns an immediate webhook response to keep the conversation seamless without timeout errors.

  • Conditional Router Logic:

    • Branch 1 (New Appointment Booked): Creates a new patient/appointment record in Airtable, sends a booking confirmation via Gmail, and responds to the frontend webhook.

    • Branch 2 (Rescheduled): Searches existing Airtable records, updates the existing record with the new appointment timestamp, and notifies the team/client via Gmail.

    • Branch 3 (Cancelled): Searches for the matching record in Airtable, updates the status to “Cancelled,” and sends an automated cancellation email.

Flow 2: Outbound Voice AI Dispatcher

To automate proactive patient reminders or lead follow-ups, I built an outbound dispatch scenario:

  1. Airtable Trigger: Watches for updated records or scheduled reminder times.

  2. HTTP Module (POST /call/phone): Sends a payload to the Voice AI API with dynamic variables (patient name, original appointment time, clinic notes) to initiate an automated phone call.

  3. Airtable Upsert: Updates the database record with the call status, duration, and disposition flags.

Key Takeaways & Technical Best Practices

  • Fast Webhook Responses: Always place a Webhook Response module near the front of your scenario when working with real-time Voice AI engines to prevent voice latency.

  • Atomic Database Operations: Use unique search keys (e.g., Phone Number or Booking ID) in Airtable search modules before updating or creating records to prevent duplicate entries.

  • Error Handling: Ensure fallbacks are configured on HTTP and Airtable modules so temporary API timeouts don’t leave customer records out of sync.

If you are building custom AI Voice Agents or need help connecting complex webhooks with CRMs/databases, feel free to ask questions below!

1 Like