This AI Voice Agent can do anything… just watch

Key Concepts

AI Voice Agents, NA10, Retail, Hostinger VPS, Webhooks, OpenAI Models, Google Calendar Integration, System Prompts, Voice Customization, Niche Specialization, Automation, Workflows, Vector.

Building AI Voice Agents: A Comprehensive Guide

Introduction

The video emphasizes the growing importance and potential of AI voice agents in various business applications. It highlights the increasing market trend and asserts that voice interaction is the future modality, surpassing text-based communication due to its speed and convenience. The video aims to guide viewers, even beginners, through the process of building and deploying their own AI voice agent.

Why Voice Agents Matter

• Market Growth: The voice AI agent market is experiencing significant growth, making it an opportune time to enter the field.
• Future Modality: Voice is positioned as the next major modality in AI, driven by its speed (three times faster than typing) and accessibility.
• Hyperrealistic Voice: Cutting-edge voice models are now highly realistic, with over 95% of people unable to distinguish them from human voices.
• Convenience: Voice interaction is convenient in various situations, such as driving, walking, or attending meetings.
• Accessibility: Voice agents cater to a broader audience, including those with limited literacy skills.
• Business Need: Every business can benefit from voice agents for lead generation, time-saving, calendar management, and automation.

Building a Voice Agent from Scratch

The video provides a step-by-step guide to building a voice agent using NA10, Retail, and Hostinger VPS.

1. Setting up Hostinger VPS

• Problem: AI agents need to run 24/7, which can be expensive with platforms like NA10 Cloud due to execution limits.
• Solution: Hostinger VPS offers a cost-effective solution with the KVM2 plan, capable of running hundreds of AI agents.
• Recommendation: Opt for the 24-month plan for long-term cost savings.
• Setup:
  • Select the server location closest to you.
  • Choose NA10 as the operating system using the one-click install template.
  • Use coupon code "David" for an additional 10% discount.
• Benefits: Hostinger VPS provides data visibility, unlike NA10 Cloud.

2. Building the Agent in NA10

• Trigger: Use the Webhook trigger to receive data.
  • Change the HTTP method to POST.
  • Set "Respond To" to "Respond Node Webhook."
  • Save the production URL for later use.
• AI Agent Node:
  • Add an AI Agent node to process the data.
  • Set the user message to pass the transcription from the call (included in the webhook).
  • Add a system prompt to control the agent's behavior.
  • Example System Prompt (for a booking agent):
    • "You are a booking agent for [Business Type]. Your goal is to help the user book, change, or cancel an appointment. Be polite and professional. If the user asks a question you cannot answer, say 'I am sorry, I cannot answer that question. Please call the business directly.'"
    • Specify the time zone or instruct the agent to ask the user for their time zone.
  • Select an OpenAI model (e.g., GPT-4.0 Turbo) and provide your API key.
• Tools (Google Calendar Integration):
  • Get Many:
    • Connect to your Google Calendar account.
    • Set the operation to "Get Many" to check for available slots.
    • Select the target calendar.
    • Let the model define the parameters.
    • Rename the node to "Get Many."
  • Create Event:
    • Connect to your Google Calendar account.
    • Set the operation to "Create" to create a new event.
    • Select the target calendar.
    • Let the model define the parameters (description, summary, etc.).
    • Rename the node to "Calendar Create Event."
• Respond to Webhook:
  • Add a "Respond to Webhook" node and leave it on "First Incoming Item."
  • Enable the automation in NA10.

3. Integrating with Retail (Voice Agent)

• Retail Setup:
  • Create a free account on Retail.
  • Create a new voice agent using the "Single Prompt Agent" template.
  • Choose a model (e.g., GPT-4.0 Turbo).
  • Set a system prompt (e.g., for a barber shop).
  • Select "AI Speaks First" and "Custom Message."
• Speech Settings:
  • Add background noise (e.g., coffee shop) for realism.
• Functions (Connecting to NA10):
  • Add a custom function named "Calendar Agent."
  • Provide a clear description (e.g., "This tool is to check the available times on the user's calendar. After giving the available times, this tool will be used again to book the available time slot.").
  • Paste the production URL from the NA10 Webhook trigger into the endpoint field.
  • Enable "Speak During Execution" and disable "Speak After Execution."
• Voice Customization:
  • Select a voice from 11Labs, OpenAI, or Cartisia.
  • Fine-tune the voice model (e.g., 11Labs Turbo V2).
  • Adjust voice speed (1.1 recommended), temperature (cooler for business), and volume.

Making Money with Voice Agents

• Niche Specialization: Focus on a specific niche (e.g., consulting, coffee shops, mechanics, barbers) that you understand.
• Pre-built Agents: Create pre-built agents tailored to the chosen niche (80% complete) and customize the remaining 20% for each client.
• Higher Perceived Value: Voice agents have a higher perceived value than chatbots.
• Automated Prompt Creation: Build an agent that creates custom prompts for new businesses.
• Upwork Opportunities: Numerous companies are seeking voice agents on platforms like Upwork, willing to pay thousands of dollars.

Vector: An AI-First Productivity App

• Workflows: Vector offers workflows, which are mini-automations within the app.
• Example: Create a workflow to generate ideas every 3 hours.
• Benefits: Vector helps users be more productive by automating tasks.

Conclusion

The video provides a comprehensive guide to building and deploying AI voice agents, emphasizing their potential for business automation and revenue generation. It encourages viewers to take action and leverage the provided resources to create their own voice agents. The combination of NA10, Retail, and Hostinger VPS offers a powerful and cost-effective solution for building and deploying these agents.