Skip to content

Latest commit

 

History

History
110 lines (83 loc) · 3.26 KB

README.md

File metadata and controls

110 lines (83 loc) · 3.26 KB

AI Call Assistant

An intelligent outbound calling system powered by VAPI AI, Make.com workflows, and Twilio infrastructure. The assistant speaks fluent English and Hindi, switching naturally mid-conversation to match the caller.

Core Architecture

VAPI AI

The conversational AI engine for outbound calls. Delivers natural language understanding, bilingual (English/Hindi) fluency, and adaptive call handling.

Make.com Workflows

Orchestrates the entire call pipeline:

  • Webhook Trigger: Receives user data from the contact form
  • Supabase Integration: Stores and retrieves contact information
  • Data Processing: Normalizes phone numbers and user data
  • Array Aggregator: Iterates through contacts for batch calling
  • HTTP Module: Triggers VAPI calls with proper authentication and payload

Twilio Integration

Provides the underlying telecommunications infrastructure:

  • Phone number provisioning
  • Call routing and management
  • Inbound/outbound call handling
  • IVR capabilities

Get a Twilio number (available on trial accounts)

System Flow

Contact Form → Node.js API → Make.com Webhook 
→ Supabase (Data Storage) → Data Processing 
→ VAPI AI Assistant → Twilio (Phone Calls)

Key Capabilities

  • Bilingual voice experiences in English and Hindi, including smooth code-switching
  • Automated outbound campaigns with batch processing and contact enrichment
  • Real-time orchestration via Make.com with Supabase-backed data
  • Carrier-grade delivery and IVR options through Twilio
  • Secure credential handling with API-keyed requests to VAPI and Twilio

Setup & Configuration

Prerequisites

  • Node.js & npm
  • VAPI AI account and API key
  • Make.com account and workflow
  • Twilio account with phone number
  • Supabase project (for data storage)

Installation

npm install
npm start

Server runs on: http://localhost:3000

Configuration Guide

1. Twilio Setup

  • Buy a phone number from Twilio Console
  • Note your phone number ID (used in Make.com workflow)

2. VAPI Configuration

  • Create an AI assistant in VAPI
  • Configure assistant behavior and language settings
  • Get your assistant ID and API key
  • Set authentication headers in Make.com HTTP module

3. Make.com Workflow

  • Webhook receives contact data
  • Supabase module stores/retrieves phone numbers
  • Array Aggregator processes batch calls
  • HTTP module calls VAPI with payload: 
    {
  "customer": {
    "number": "{{phone_number}}",
    "name": "{{user_name}}",
    "extension": ""
  },
  "phoneNumberId": "<your_twilio_id>",
  "assistant": {{assistant_config}}
}

4. Server Integration

Update /api/call endpoint in server.js to connect with your Make.com webhook

Technologies

  • Backend: Node.js, Express.js
  • AI: VAPI AI
  • Workflow Automation: Make.com
  • Telecommunications: Twilio
  • Database: Supabase
  • Frontend: Vanilla HTML/CSS/JavaScript

Screenshots

  • Call outcome preview:
    AI Call Proof
  • Make.com workflow orchestration: Make.com Workflow

Feel free to reach out to me in case you have any queries!