A production AI voice agent built in Python for an automotive service company.
The agent answers inbound calls, qualifies leads, books appointments, sends SMS/email confirmations, and logs customer data to Google Sheets, enabling the business to provide 24/7 receptionist coverage without additional staff.
The initial implementation was built using Vapi. However, after testing in a production environment, the conversation quality, responsiveness, and workflow flexibility did not consistently meet the project's requirements.
To achieve better control over the voice experience and business logic, the system was rebuilt using Deepgram's Voice Agent stack.
- Answer inbound calls automatically
- Qualify leads and collect customer information
- Schedule appointments through Google Calendar
- Send SMS and email confirmations
- Log customer interactions to Google Sheets
- 24/7 availability
- Python
- Deepgram Voice Agent
- Twilio
- Google Calendar API
- Google Sheets API
- SMS & Email Automation
This repository contains a safe demonstration of the architecture used in production. All credentials, client-specific data, and sensitive configurations have been removed before publication.