This AI Voice Agent Can Handle EVERYTHING! (No-Code Tutorial)

Key Concepts

• No-Code AI Agent: Building an AI agent without writing code.
• Inbound Voice AI Agent: An AI agent that answers incoming phone calls.
• Retail AI: A no-code platform for building voice AI agents.
• Naden: A no-code automation platform.
• Persona: Defining the AI agent's role, skills, and objectives.
• Knowledge Base: Providing the AI agent with relevant information.
• Dynamic Variables: Placeholders for information that changes, like caller details.
• Functions: Pre-built actions the AI agent can perform, like checking calendar availability.
• Webhooks: A way for applications to communicate with each other in real-time.
• Structured Output Parser: A tool to define the format of the AI agent's output.
• CRM (Customer Relationship Management): A system for managing customer interactions and data.

Building a No-Code Inbound Voice AI Agent

Problem Statement

• Businesses miss a significant percentage (20-40%) of inbound calls, leading to lost revenue.
• Dental clinics miss 27% of calls, auto shops miss 20%, and banks miss nearly half.
• Missed calls often represent leads ready to convert.

Solution

• An AI agent that answers calls, qualifies leads, collects information, checks insurance, and books appointments.
• The agent can update a CRM automatically after the call.

Demo Overview

• The video demonstrates an AI agent named "Katie" answering a call to Crescent Family Dental.
• Katie greets the caller, asks how she can assist, and confirms that the clinic accepts the caller's insurance (Sigma).
• Katie books an appointment for a dental cleaning on Monday, June 30th, 2025, at 12 p.m.
• Katie collects the caller's name (John Smith), phone number (510-223-1125), and email address ([email protected]).
• The information is then sent to a Naden setup, which updates a Google Sheet (acting as a CRM) and sends a confirmation email.
• The appointment is automatically booked in Google Calendar, including a Google Meet link.

Tools Used

• Retail AI: For building the voice AI agent.
• Naden: For automating the post-call workflow (CRM update, email sending).

Easiest Setup Method (Using Blueprints)

1. Log into Retail AI and Naden.
2. Join the creator's community and download the Retail AI and Naden blueprints.
3. In Retail AI, import the Retail blueprint.
4. In Naden, import the Naden blueprint.
5. Customize the imported templates as needed.

Step-by-Step Build (From Scratch)

Retail AI Setup

1. Create a Voice Agent: In Retail AI, create a new agent and select "Single Prompt Agent."
2. Configure Settings: Leave most settings as default. Change the model to "GPT-4 Turbo" and select "Fast" for a more natural voice.
3. Define the Prompt:
   • Persona: Define the AI agent's name (e.g., Katie), role (receptionist), skills, and objectives.
   • Knowledge Base: Provide the agent with business information, accepted insurance providers, and current time (using dynamic variables).
   • Caller Information: Specify the information to collect (first name, last name, phone number, email, insurance provider).
   • Rules: Set guidelines for the agent's communication style (simple, clear, friendly).
   • Steps to Follow: Outline the call flow (greet the caller, understand the reason for the call, collect information if booking an appointment).
4. Add Functions:
   • Check Calendar Availability: Use the cal.com integration. Requires a cal.com account, API key, and event type ID.
   • Book Appointment: Similar to "Check Calendar Availability," using cal.com.
   • End Call: Ends the call.
   • Transfer Call: Transfers the call to a specified phone number. Supports cold and warm transfers.
5. Web Hook Settings: Copy the test URL from Naden and paste it into the Web Hook settings in Retail AI.

Naden Setup

1. Add a Web Hook: Add a Web Hook node to the Naden workflow.
   • Change the method to "POST."
   • Copy the test URL.
   • Change the path to something descriptive (e.g., "inbound").
2. Add AI Agent: Add an AI Agent node.
   • Select an OpenAI chat model (e.g., GPT-4 Turbo).
   • Define the prompt: "Summarize the following transcript... extract name, email, phone number, insurance provider in JSON format."
   • Use the transcript from the Web Hook as input.
3. Require Specific Output Format: Enable this option in the AI Agent node.
4. Structured Output Parser: Add a Structured Output Parser to define the output format (call summary, first name, last name, phone, email, insurance).
5. Add Google Sheets Integration: Add a Google Sheets node to update a Google Sheet.
   • Select the Google Sheet.
   • Choose the "Update Row" operation.
   • Map the output from the AI Agent node to the corresponding columns in the Google Sheet.
   • Specify a column to match on (e.g., first name).
6. Optional Integrations: Add nodes for Hubspot CRM, email, Slack, or other services.

Key Arguments and Perspectives

• Automation is Crucial: Automating inbound call handling is essential for businesses to avoid missing leads and losing revenue.
• No-Code Tools Empower Users: No-code platforms like Retail AI and Naden make it easy for anyone to build and deploy AI-powered solutions.
• Customization is Key: The AI agent's persona, knowledge base, and workflow can be customized to fit the specific needs of different industries.

Notable Quotes

• "Most businesses miss 20 to 40% of their inbound calls and those miss calls are usually leads ready to convert."
• "This ai agent doesn't just answer it listens qualifies collects information checks things like in insurance and even updates your crm automatically after the call"

Technical Terms and Concepts

• Single Prompt Agent: An AI agent that uses a single prompt to guide the conversation.
• Dynamic Variables: Placeholders for information that changes, like caller details.
• Functions: Pre-built actions the AI agent can perform, like checking calendar availability.
• Webhooks: A way for applications to communicate with each other in real-time.
• Structured Output Parser: A tool to define the format of the AI agent's output.

Logical Connections

• The Retail AI agent handles the voice interaction and collects data.
• The Naden workflow receives the data from Retail AI via a webhook.
• Naden uses an AI agent to summarize the call and extract key information.
• Naden then updates a Google Sheet (CRM) and can trigger other actions (email, Slack, etc.).

Data and Statistics

• Unanswered call rates by industry (from Invoca):
  • Auto dealerships: 37%
  • Auto service centers: 39%
  • Banks: 51%
  • Dental offices: 37%

Synthesis/Conclusion

The video provides a comprehensive guide to building a no-code inbound voice AI agent using Retail AI and Naden. By defining a persona, providing a knowledge base, and using pre-built functions, users can create an AI agent that answers calls, qualifies leads, and automates tasks like appointment booking and CRM updates. The use of no-code tools makes this technology accessible to a wider audience, enabling businesses to improve their customer service and capture more leads. The video emphasizes the importance of customization and provides resources for further learning and collaboration.