Build a Real Voice AI System with Antigravity + n8n

Full Stack Voice AI System Setup – Detailed Summary

Key Concepts:

• VoiceAI: Utilizing AI to automate phone calls for business purposes (appointment booking, lead qualification, etc.).
• Anti-Gravity: A Google tool for rapid website construction, used here as the front-end interface.
• NAN (Noodl): A no-code automation platform serving as the back-end logic processor.
• Retail AI: A voice agent platform providing the AI voice and conversational capabilities.
• Webhook: A method for real-time data transfer between Anti-Gravity (front-end) and NAN (back-end).
• Cal.com: A scheduling platform integrated with Retail AI to check calendar availability and book appointments.
• API Key: A unique identifier used to authenticate access to services like Retail AI and Cal.com.
• JSON (JavaScript Object Notation): A standard data format used for transmitting data between systems.

---

I. System Overview & Demonstration

The video demonstrates a fully functional VoiceAI system capable of autonomously handling appointment booking. The system initiates with a user submitting a form on a website. Upon submission, an AI assistant (Cindy, powered by Retail AI) calls the user, confirms details, and books an appointment based on calendar availability. The entire system is built using a combination of no-code/low-code tools: Anti-Gravity for the front-end website, NAN for the automation back-end, and Retail AI for the voice agent. This system is presented as a viable solution for businesses like massage therapists, electricians, auto shops, and other local contractors who often miss calls.

The initial demo showcases a successful booking: a massage appointment scheduled for Zubar on Wednesday, the 28th at 12:30 PM, confirmed via email (zubar.trabzaggmail.com) and phone number (512331-2211). The system accurately checked calendar availability, offered alternative times when the initial request was unavailable, and completed the booking process. The organizer's calendar (the creator's) is the only one shown due to email connection. A 30-minute meeting invite was successfully created.

II. Building the Front-End with Anti-Gravity

The first step in building the system involves creating the front-end website using Anti-Gravity. A pre-built prompt (available in the video description) is used to generate a modern, calm, and premium one-page website for "Harbor Wellness Massage Center" in San Francisco. The prompt specifies the desired look and feel, including hours of operation and a relaxing, professional aesthetic.

Key Prompt Elements:

• Website type: Modern, calm, premium one-page website.
• Business: San Francisco massage therapy business ("Harbor Wellness Massage Center").
• Desired aesthetic: Relaxing, professional, clean.

The Anti-Gravity process involves pasting the prompt into the tool and initiating website construction. A crucial element is the webhook URL embedded within the prompt. This URL is the communication pathway to the NAN back-end, transmitting form submission data. The presenter emphasizes the importance of installing the Google Chrome extension for Anti-Gravity to monitor the website building process.

III. Setting Up the Back-End with NAN

The back-end is constructed using NAN. The process begins with a blank workflow. A webhook trigger is added, configured to listen for POST requests from the Anti-Gravity front-end.

Webhook Configuration:

• HTTP Request Method: POST
• Path: A customizable path (e.g., "Gemini" or "Anti-Gravity") used for identification.
• Immediately Trigger: Enabled for real-time response.

The webhook receives data (name, email, phone number, service requested) from the submitted form. This data is then "pinned" within NAN, making it accessible to subsequent nodes in the workflow.

Next, an HTTP request node is added to communicate with the Retail AI API.

HTTP Request Node Configuration:

• HTTP Request Method: POST
• URL: api.ai.com/v2/create_phone_call (Retail AI API endpoint)
• Authentication: Generic credential type, custom header with "Authorization" set to "Bearer" followed by the Retail AI API key. "Content-Type" is also included.
• Body: A JSON payload (provided in the video description) containing the necessary data for Retail AI to initiate the call and handle the conversation. This JSON payload includes dynamic variables that will be populated with the data received from the Anti-Gravity form.

IV. Configuring the Voice Agent with Retail AI

Retail AI is used as the voice agent. The presenter highlights Retail AI's popularity and recent recognition as a fast-growing startup.

Retail AI Configuration:

• Agent Creation: A "Single Prompt Agent" is created.
• Model Selection: GPT Realtime Mini is recommended for low latency.
• Prompt: A detailed prompt (available in the video description) defines the agent's persona ("Cindy, the AI powered voice assistant for Harbor Wellness Massage"), rules of engagement, and specific conversational flows.
• Agent ID Override: The Retail AI Agent ID (found in the Retail AI settings) must be copied and pasted into the corresponding field in the Retail AI prompt.
• Function Integration: Retail AI integrates with Cal.com for calendar availability checks and appointment booking. This requires adding two functions:
  • Check Calendar Availability: Uses the Cal.com API key and event type ID.
  • Book Calendar: Uses the Cal.com API key and event type ID.

Cal.com Integration:

• An API key is generated from the Cal.com settings.
• The Event Type ID is obtained from the specific event type created within Cal.com.
• Both the API key and Event Type ID are added to the corresponding functions in Retail AI.

V. Testing and Deployment

After configuring all components, the system is tested by submitting a form on the Anti-Gravity website. This triggers the webhook, sends data to NAN, which then makes an API call to Retail AI, initiating the phone call. The presenter demonstrates the successful flow, confirming that the system accurately processes the request and books the appointment.

VI. Community Resources & Conclusion

The presenter directs viewers to several resources:

• Google Doc with Prompts: Contains the Anti-Gravity prompt, Retail AI prompt, and NAN JSON. (Link in description)
• Community Classroom: Offers access to pre-built workflows and prompts for VoiceAI projects.
• AI Agency Course: A five-week program for building an AI agency.
• Free Community: Provides access to free resources and a supportive community.

The video concludes by encouraging viewers to like and subscribe, emphasizing the potential of VoiceAI and the available resources for learning and implementation. The core takeaway is that a fully functional, automated appointment booking system can be built using readily available no-code/low-code tools, offering a valuable solution for businesses seeking to improve lead capture and customer service.