This Voice AI Calls Your Leads Instantly – Watch It in Action!

Key Concepts

• Voice AI Agent: An automated system using artificial intelligence to conduct phone conversations.
• Lead Qualification: The process of determining if a potential customer is a good fit for a product or service.
• No-Code Workflow: A process built using visual interfaces and pre-built components, requiring minimal or no coding.
• NAND: A no-code platform used to automate workflows and integrate different applications.
• Retail AI: A platform providing voice AI agent services.
• HTTP Request Node: A component in NAND used to interact with APIs.
• Web Hook: A mechanism for applications to send real-time information to each other when a specific event occurs.
• Summarization Chain: A process that uses AI to condense a large amount of text into a shorter summary.
• API (Application Programming Interface): A set of rules and specifications that software programs can follow to communicate with each other.
• JSON (JavaScript Object Notation): A lightweight data-interchange format that is easy for humans to read and write and easy for machines to parse and generate.

Building a Voice AI Agent Workflow with NAND and Retail AI

Overview

The video demonstrates how to build a no-code workflow using NAND and Retail AI to automatically call leads who submit a form, qualify them using a voice AI agent, and update their information in a Google Sheet. The goal is to improve sales conversions by contacting potential customers immediately after they express interest.

Step-by-Step Process

1. Triggering the Workflow:

   • Use the "On Form Submission" trigger in NAND.
   • Create a form within NAND or integrate with an existing form (e.g., from a website).
   • Include fields like name, email, phone number, budget, and a description of the service needed.
   • Pin the data from the form submission to avoid re-entering it during testing.

2. Initiating the Voice Call with Retail AI:

   • Use an "HTTP Request" node to interact with the Retail AI API.
   • Set the method to "POST" and the URL to the Retail AI "create call" endpoint.
   • Use "Generic Credential Type" with "Custom Oauth2" for authentication, providing your Retail AI API key in the headers ("Authorization: Bearer YOUR_API_KEY").
   • Set the body content type to "application/json" and specify the body using JSON.
   • Include parameters like "from number" (your Retail AI phone number), "to number" (customer's phone number from the form), "name," "email," "service," "budget," and "override AI agent ID."

3. Configuring the Retail AI Agent:

   • Create an agent in Retail AI, selecting a "Single Prompt Agent" and starting from blank.
   • Choose a voice model (e.g., GPT 4.1) and a voice (e.g., Anthony or a custom voice).
   • Write a universal prompt for the agent, defining its role, conversational style, and objective.
     • Example prompt: "You are Jarvis, an AI-powered voice assistant for AI Workshop AI agency. Your role is to qualify leads after they submit a form."
     • Include dynamic variables (e.g., {{name}}, {{phone number}}) to personalize the conversation with data from the form submission.
   • Add functions to the agent, such as "end call" and "transfer call."
     • "End call" function: Simply hangs up the phone.
     • "Transfer call" function: Transfers the call to a human agent's phone number.
   • Configure web hook settings to send the conversation transcript back to NAND after the call is complete.

4. Updating the Google Sheet:

   • Use a "Google Sheets" node to append or update a row in a Google Sheet.
   • Set up credentials to connect to your Google account and select the appropriate spreadsheet and sheet.
   • Match the row based on the phone number from the form submission.
   • Populate the name, email, phone number, and budget columns with data from the form submission.
   • Leave the AI phone summary column blank for now.

5. Summarizing the Conversation and Updating the Google Sheet (Web Hook Workflow):

   • Use a "Web Hook" node to receive the conversation transcript from Retail AI.
   • Set the method to "POST" and the path to a unique identifier (e.g., "/lead").
   • Use a "Summarization Chain" node to summarize the transcript.
     • Select an OpenAI model (e.g., GPT-4 mini).
     • Set the summarization method and prompt.
     • Replace the default text in the prompt with JSON.transcript to use the transcript from the web hook.
   • Use another "Google Sheets" node to update the same row in the Google Sheet.
   • Match the row based on the phone number.
   • Populate the AI phone summary column with the summary from the summarization chain.

Key Arguments and Perspectives

• Importance of Immediate Contact: Sales conversions are significantly higher (391%) when customers are contacted within the first minute of submitting a form.
• Transparency with AI: It's better to be upfront about using an AI voice agent rather than pretending it's a human.
• Humor and Engagement: Injecting humor into the AI agent's script can increase engagement and make the interaction more user-friendly.
• Flexibility and Customization: The workflow can be customized with features like call settings, keypad detection, and background noise.

Notable Quotes

• "Sales conversions are 391% higher in the first minute if somebody calls the customer as soon as they submit a form."
• "The ideal time to contact a potential customer is exactly when they enter their phone number into a form." - Carl Norman, Business Development Director at Leak Connect
• "You want to make sure that you're letting the customer know ahead of time that this is a voice AI agent but you want to have a humor so that way it uh keeps that interaction very engaging."

Technical Terms and Concepts

• Dynamic Variables: Placeholders in the AI agent's script that are replaced with data from the form submission (e.g., {{name}}, {{phone number}}).
• Functions: Capabilities that can be added to the AI agent, such as ending the call or transferring it to a human.
• Override AI Agent ID: The unique identifier of the Retail AI agent that you want to use for the call.
• From Number: The phone number that the AI agent will use to make the call (must be purchased through Retail AI/Twilio).

Logical Connections

• The form submission triggers the workflow, sending data to Retail AI.
• Retail AI initiates the voice call and follows the script defined in the agent's prompt.
• After the call, Retail AI sends the transcript to NAND via a web hook.
• NAND summarizes the transcript and updates the Google Sheet with the summary.

Data and Statistics

• Sales conversions are 391% higher when customers are contacted within the first minute.

Synthesis/Conclusion

The video provides a detailed guide to building a voice AI agent workflow that automates lead qualification and improves sales conversions. By combining the no-code capabilities of NAND with the voice AI features of Retail AI, businesses can create a system that contacts leads immediately, qualifies them using a customizable AI agent, and updates their information in a CRM (in this case, a Google Sheet). The key takeaways are the importance of immediate contact, transparency with AI, and the potential for humor and customization to enhance the user experience. This workflow can be adapted and sold to local businesses or used to improve internal processes.