Full-Stack Voice AI Agent with LiveKit, n8n and MCP on AWS

Build and deploy a real-time Voice AI Agent using LiveKit, n8n, MCP, and AWS with full automation and integration.

What you'll learn
- Build and deploy a full-stack Voice AI Agent using LiveKit, n8n, and MCP on AWS.
- Set up and configure an Ubuntu server on AWS EC2 for development.
- Create and manage virtual environments for Python projects.
- Integrate LiveKit for real-time voice communication in AI systems.
- Configure and connect OpenAI and Deepgram APIs for voice interaction.
- Build and run the AI Agent, and test it via the LiveKit Playground.
- Integrate prompts. py with the AI Agent for task and session control.
- Add and test AI Avatars using Tavus for realistic voice experiences.
- Deploy and manage the AI Agent as a system service using Systemctl.
- Build and deploy a React/Next.js frontend and connect it to the backend.
- Secure your AI application using NGINX and Let’s Encrypt SSL.
- Automate booking workflows and email notifications with n8n.
- Integrate Google Calendar and Gmail nodes within n8n workflows.
- Configure PM2 to manage and monitor Next.js production deployments.
- Test, verify, and troubleshoot AI booking and automation flows end to end.

Requirements
- Basic understanding of Python programming and command-line usage.
- A free or paid AWS account to create and manage EC2 instances.
- Basic knowledge of AI tools or APIs like OpenAI and Deepgram (optional).
- Enthusiasm to learn, explore, and build a real-time Voice AI system from scratch — beginners are welcome!

Description
Build your ownend-to-end Voice AI AgentusingLiveKit,n8n, andMCP— hosted onAWS.This course guides you step by step in developing areal-time intelligent Voice AI systemwith automation, voice interaction, and web integration.

You’ll begin by setting up and configuring anUbuntu server on AWS EC2, preparing yourPython environment, and integratingLiveKitfor real-time communication. Then, you’ll connect your AI Agent withn8nfor workflow automation andMCPfor multi-channel task handling — enabling features likevoice-based appointment booking, email notifications, andcalendar scheduling.

Once the backend is complete, you’ll build and deploy aReact/Next.js frontend, secure it usingNGINXandSSL certificates, and manage your application usingPM2andSystemctl.By the end of the course, you’ll have afully functional, secure, and automated Voice AI Agentdeployed in a cloud environment.

Introduction

What You’ll Learn

System Setup on AWS EC2

Set Up and Configure an Ubuntu Server on AWS EC2

Connect to Your AWS EC2 Instance Using MobaXterm

Set Up the Project Directory

Set Up Python Virtual Env

LiveKit Essentials: Setup, API Keys, and Configuration

What is Livekit?

Why LiveKit for Our AI Project?

LiveKit Documentation Overview

Installing LiveKit and Its Dependencies

Create LiveKit Account

Set Up Your OpenAI API Key

Deepgram Account & API Setup

LiveKit Plugins Setup

Build and Test Your AI Agent in LiveKit

Build and Run the Agent Script (agent)

Access via LiveKit Playground

Integrate prompts. py with AI Agent

How prompts. py Works in the Project

Add Task Definition

Define Session Instruction

Update agent. py Based on prompts. py

Run Voice AI Agent

Integrating AI Avatars with Your Voice AI Agent

What is an AI Avatar?

How AI Avatar Integration Works

Tavus Account & Persona Setup

Add Tavus Persona IDs and APIs to .env

Add Avatar in agent. py

Rerun Voice AI Agent

Test AI Avatar via LiveKit Playground

Configure AI Agent as a Systemctl Service

Create a Systemctl Service File for the Voice AI Agent

Integrating LiveKit AI Agent with a Custom React Frontend

Integrate with React Frontend

Clone the React Project and Install Node.js

Run the React App in Development Mode

Access the React Frontend via EC2 Public IP

Deploy Next.js Frontend to Production Using PM2

Overview of Next. js Production Deployment

Build & Fix Build Issues

Run with PM2

Verfiy deployment

Purchase a Domain and Configure DNS Records

Purchase a Domain from Godaddy

Configure A Records

Secure the Application using NGINX and Let's Encrypt SSL

Overview of the Project

Set Up and Configure NGINX

Install Certbot Let’s Encrypt to Enable HTTPS

Verify HTTPS Access

Auto Renew SSL Certificates

Integrating AI Agent with n8n and MCP

Overview: AI Agent with n8n and MCP

Overview of n8n

What is MCP?

How It Works: AI Agent - MCP - n8n Flow

Modify Agent Configuration for MCP Server

Add an MCP Server Trigger Node in n8n

Add a Google Calendar Node in n8n

Fix Google Calendar Access

Add Another Google Cal Node

Add MCP Server URL in .env File on Server

Restart Agent and Verify n8n Integration

Modify prompts. py for Appointment Booking Flow

Book an Appointment with Voice AI Agent

Booking Validation: Prevent Double Appointments

AI Agent: Book the Appointment and Send Email Notification

Add a Gmail Node in n8n workflow

Set the Parameters in Gmail Node in n8n

Modify prompts. py for Email Notifications

Test the Voice AI Agent: Book Appointment and Send Email Notification

Last Lecture

Who this course is for:
- Developers and AI enthusiasts who want to build real-time Voice AI applications from scratch.
- Full-stack engineers looking to integrate voice automation, workflow orchestration, and AI services.
- Students or professionals interested in learning practical cloud deployment and automation on AWS.
- Anyone eager to explore how LiveKit, n8n, and MCP work together to create an end-to-end Voice AI system.
More Info