This n8n Voice AI Agent CALLS Your Customers (No Humans Needed!)

Key Concepts

Voice AI Agents: Automated systems that use voice to interact with users, mimicking human conversation.
Retail AI: A platform for building and deploying voice AI agents.
Naden.io: A no-code platform for creating automated workflows.
Universal Prompt: The core instruction set for a voice AI agent, defining its persona and behavior.
Dynamic Variables: Placeholders in prompts that are replaced with real-time data during execution.
Webhooks: Mechanisms for enabling communication between different applications or services.
API Key: A unique identifier used to authenticate and authorize access to an API.
HTTP Request Node: A component in Naden.io that allows interaction with external APIs.
Knowledge Base: A repository of information that a voice AI agent can access to provide more contextually relevant responses.
Call Transfer (Cold/Warm): Functionality allowing the AI agent to transfer a call to a human agent, with or without providing context.
GPT Realtime Mini: A fast and efficient AI model for voice agent interactions, though potentially expensive.
Latency: The delay between an input and its corresponding output in a system, crucial for natural-sounding voice interactions.

Building a Lead Qualification Voice AI Agent

This guide details the process of creating a lead qualification voice AI agent using Retail AI for the front-end voice interaction and Naden.io for the back-end workflow automation. The goal is to automate the initial contact with potential clients who have expressed interest in AI solutions, gather essential information, and qualify them for further engagement.

Demo and Workflow Overview

The video begins with a demonstration of a voice AI agent, "Cindy," from "AI Workshop AI Agency," calling a prospect who had previously filled out a form about AI solutions. Cindy professionally gathers information such as the prospect's urgency to start, budget ($5,000-$10,000), and preferred follow-up time (Monday, 9:00 AM - 11:00 AM). The agent handles interruptions gracefully and concludes the call by confirming the next steps. The presenter emphasizes the naturalness of the interaction, attributing it to the use of GPT in real-time.

The core components of this system are:

Retail AI: For creating the voice agent and its conversational logic.
Naden.io: For orchestrating the workflow, including form submissions, API calls, and data storage.

Building the Voice Agent in Retail AI

Account Setup: Sign up for free accounts on both Retail AI and Naden.io.
Agent Creation:
- Navigate to the "Agents" section in Retail AI.
- Click "Create an agent."
- Select "Voice agent" and then "Single prompt agent."
- Choose "Start from blank" and click "Create."
Universal Prompt:
- This is the most critical part, defining the agent's persona and script.
- A pre-written prompt is provided (link in description).
- Key elements of the prompt:
  - Persona: "You're Cindy, an AI powered voice assistant for AI workshop AI agency." (This can be customized for your business).
  - Script for Requalification Call: A step-by-step guide for the conversation.
  - Dynamic Variables: Placeholders like {{name}} which are populated with data from the form submission (e.g., customer's name).
  - Scenario Handling: Includes logic for identifying the recipient, checking availability, and leaving messages if necessary.
  - Purpose: Clearly states the reason for the call: "You recently filled out a form about AI solutions. I'm here to ensure you're connected with the right specialist."
  - Information Gathering: Assesses urgency, budget, and other relevant details.
  - Conditional Logic: Guides the conversation based on the prospect's responses (e.g., if budget is confirmed, move to next steps).
  - Optional Functions: The prompt can be extended to include appointment booking or call transfer capabilities.
- Saving the Prompt: Copy the entire prompt and paste it into the Retail AI agent's prompt section. Click "Save."

Building the Workflow in Naden.io

Account Setup: Sign up for a free account on Naden.io.
Workflow Creation:
- Naden.io uses a visual canvas for building workflows.
- Trigger: The workflow starts with a trigger. In this case, a "Form Submission" is used.
  - Naden Form: A built-in form can be created within Naden.io.
  - Webhook: Alternatively, a webhook can be used to connect to a form on your own website.
- Form Elements: Define the fields for the form:
  - Name (Text, Required)
  - Email (Email, Required)
  - Phone (Number, Required)
  - Budget (Dropdown with options like "$2k-$5k", "$5k-$10k", "$10k-$20k")
  - Service (Text Area for description)
- HTTP Request Node: This is the core node for interacting with external APIs.
  - Purpose: To send the collected form data to the Retail AI API to initiate the voice call.
  - Method: Set to "POST."
  - URL: The Retail AI endpoint for creating a phone call (found in Retail AI's API documentation).
  - Authentication:
    - Obtain your API key from Retail AI (Settings > Limits > API Key).
    - In Naden.io's HTTP request node, create a new credential using your API key.
    - Set "Content-Type" to "JSON."
  - Body:
    - Toggle "Send body" on.
    - Use "JSON" as the format.
    - Key Parameters:
      - from_number: The phone number purchased from Retail AI (or your connected number).
      - to_number: The phone number submitted by the user in the form. This is a dynamic variable from the form submission.
      - agent_id: Your Retail AI agent ID (found in the top corner of your Retail AI dashboard).
      - retail_dynamic_variable: This refers to the dynamic variables defined in your Retail AI prompt (e.g., {{name}}, {{phone}}, {{budget}}). These are mapped from the Naden.io form submission.
- Google Sheets Integration:
  - Purpose: To store the collected lead information for CRM or record-keeping.
  - Node: Use the "Sheets" node and select "Append or Update."
  - Credentials: Connect your Google account to Naden.io.
  - Sheet Selection: Choose your Google Sheet and the specific sheet tab.
  - Column Mapping: Map the fields from your form submission (Name, Email, Phone, Budget, Service) to the corresponding columns in your Google Sheet.
- Optional: Webhook for Call Summary:
  - Purpose: To receive a summary of the voice agent's conversation after the call ends.
  - Retail AI Webhook Settings: Configure Retail AI to send webhook data to a specific URL provided by Naden.io.
  - Naden.io Webhook Node: Set up a webhook node in Naden.io to receive this data.
  - Summarization Chain: Use a summarization chain (e.g., with GPT) to process the call transcript.
  - Update Google Sheet: Append the summarized information to a dedicated tab in your Google Sheet.

Customization and Advanced Features

AI Models:
- Retail AI offers various AI models. "GPT Realtime Mini" is recommended for natural interactions due to low latency (around 495ms).
- Other models like "GPT5 Mini" or "GPT4.1" are cheaper but have higher latency (e.g., 1295ms - 1675ms), which can make the interaction less natural.
- Cost: Be aware that real-time models can be expensive (e.g., 12 cents per minute for GPT Realtime Mini).
Voice Selection:
- The choice of voice depends on the AI model used. OpenAI models are limited to OpenAI voices.
- Using models like Claude allows for voices from 11 Labs, Cartisia, or OpenAI. Cartisia offers high-quality voice models.
Language: Select the appropriate language for the agent.
Functions:
- End Call: Automatically included, allows the agent to conclude the conversation.
- Call Transfer:
  - Cold Transfer: Immediately transfers the call to a human agent without context.
  - Warm Transfer: Allows for providing context to the human agent before transferring.
  - To add, click "Add" and select "Call Transfer" or "End Call." You can specify a phone number for the transfer.
- Appointment Booking:
  - Integrate with scheduling tools like Calendly or Cal.com.
  - This requires additional setup and tutorials are available on the presenter's channel.
Knowledge Base:
- Purpose: To provide the voice agent with additional context and information.
- Setup:
  - Click "Add" under "Knowledge Base."
  - Upload documents, websites, or files.
  - The prompt can then refer to this knowledge base (e.g., "If information is available in the knowledge base, here's what I found...").
  - The presenter uses an "AI Workshop FAQ" document as an example.
Welcome Message: Configure whether the AI speaks first upon answering.

Key Arguments and Perspectives

Power of Voice AI Agents: Voice AI agents are becoming increasingly sophisticated, almost indistinguishable from human agents, making them powerful tools for businesses.
Opportunity for Freelancers/Agencies: This technology presents a significant opportunity for freelancers and AI agencies to acquire clients quickly. Examples are given of individuals closing deals for voice AI agents shortly after learning the technology.
Ease of Implementation: Platforms like Retail AI and Naden.io offer no-code solutions, making it accessible for individuals without extensive programming knowledge to build these agents.
Cost vs. Naturalness: There's a trade-off between the cost of AI models and the naturalness of the voice interaction. Real-time, low-latency models provide a better user experience but are more expensive.
Scalability and Efficiency: Automating lead qualification through voice AI agents can significantly improve efficiency and scalability for businesses.

Data, Research Findings, and Statistics

Limbus: Closed a $22,000 voice AI bid.
Adam Sand: Sold three voice AI agents in 24 hours for the roofing industry.
Retail AI Phone Number Cost: Approximately $2.
GPT Realtime Mini Cost: Approximately 12 cents per minute.
Latency Figures:
- GPT Realtime Mini: ~495 milliseconds.
- GPT4.1: ~1295-1675 milliseconds.

Conclusion and Takeaways

The video provides a comprehensive, step-by-step guide to building a functional lead qualification voice AI agent. The key takeaway is that by combining Retail AI for voice interaction and Naden.io for workflow automation, businesses and individuals can create powerful, automated lead generation systems. The emphasis is on practical implementation, with readily available prompts and clear instructions for integrating various components. The presenter also highlights the growing market for voice AI services and encourages viewers to explore their community and courses for further learning and certification. The system is designed to be flexible, allowing for customization of prompts, AI models, voices, and the integration of advanced features like appointment booking and knowledge bases.