Add ChatGPT To Your Website In Minutes (n8n AI Chatbot)

AI Chatbot with NADN: Detailed Summary

Key Concepts:

AI Chatbot
NADN (No-Code AI Development Network)
AI Agent
Chat Model
Memory (Window Buffer Memory, RAG System)
Tools (Invoice Generator, etc.)
Workflows (Main Workflow, Sub-Workflow)
Webhooks
System Prompt
Website Embedding (HTML, JavaScript)

1. Introduction: Building an AI Chatbot with NADN

The video demonstrates how to build an AI chatbot using NADN, emphasizing its speed and ease of implementation. The chatbot can perform various functions, including answering queries, sending invoices, and booking appointments. The example showcases a chatbot embedded on a website that can generate and send an invoice to a user upon request.

2. Chatbot Functionality Demonstration

Basic Interaction: The chatbot is shown responding to a generic "hey" message and requesting the user's name and email.
Calendar Booking: The chatbot provides a calendar link when asked to "talk to your team."
Invoice Generation: The user requests an invoice, and the chatbot confirms that it has been sent. An invoice is then received in the user's email, demonstrating the chatbot's ability to generate and send contracts.

3. Workflow Overview

The chatbot functionality is achieved through two main workflows:

AI Chatbot Workflow (Main Workflow): This workflow handles the initial chat interaction and uses a sub-agent to call another workflow for specific tasks.
Invoice Generation Workflow (Sub-Workflow): This workflow is triggered by the main workflow to generate and send invoices.

4. Embedding the Chatbot on a Website

Public Availability: The chat must be publicly available to be embedded.
Embedded Chat Mode: Select the "embedded chat" mode in NADN.
Installation Script: Copy the provided JavaScript script from NADN.
Website Builder Integration: Use a website builder (e.g., WordPress, Wix, Go High Level) and add a "code" or "HTML" element.
Script Pasting: Paste the copied script into the code element.
Production URL: Replace the placeholder production URL in the script with the actual URL from NADN.
Activation: Save the changes and mark the chat as active in NADN.

Example: The video uses Go High Level to demonstrate the embedding process, but the steps are generally applicable to other website builders.

5. AI Agent Configuration

AI Agent Addition: Add an AI agent to the workflow to handle customer queries.
Chat Model Selection: Choose a chat model (e.g., OpenAI).
Memory Selection: Select a memory type. The video uses "Window Buffer Memory" for simplicity. A "RAG (Retrieval-Augmented Generation) system" is recommended for more advanced applications.
- RAG System: Described as an advanced employee trained on all company data, capable of finding relevant answers to queries.
Tool Addition: Add tools to extend the chatbot's functionality (e.g., invoice generation, image creation, appointment booking).

6. Invoice Generation Workflow Details

Trigger: The invoice generation workflow is triggered by the main AI chatbot workflow.
Call Workflow Tool: Use the "Call Workflow" tool in the main workflow to trigger the invoice generation workflow.
Tool Description: Provide a description for the tool to guide the AI agent on when to use it (e.g., "Call this tool when the user wants to generate an invoice").
Parameters: Define parameters needed for the tool, such as the user's name and email.
AI-Powered Data Extraction: Use AI to extract the user's name and email from the chat.
- Example: {{? fromAI key="name" description="This is the user's name"}}
HTTP Request: Send the extracted data to a third-party service (e.g., Go High Level) using an HTTP request.
Webhook URL: Include the execution.resumeURL in the HTTP request to allow the workflow to resume after the invoice is generated.

7. Go High Level Integration (Invoice Generation Example)

Inbound Webhook: Receive the data from NADN in Go High Level using an inbound webhook.
Contact Creation: Create a contact in Go High Level using the received data.
Invoice Creation: Generate an invoice in Go High Level.
Webhook Response: Send a webhook back to NADN with the invoice details, using the execution.resumeURL.

8. System Prompt Configuration

System Message: Add a system message to the AI agent to provide context and instructions.
Tool Section: Define the available tools and their descriptions.
Rules Section: Specify rules for the AI agent to follow (e.g., how to respond to user requests, when to use specific tools).
Conversation Flow: Guide the AI agent on how to start and maintain the conversation.

Example System Prompt Snippet:

You are an AI chatbot assisting users on a website. Follow these instructions carefully.

**Tools:**

*   **Invoice Generator:** You must have the user's email and name before generating the invoice. Only call this tool if the user asks for you to send it.

**Rules:**

*   After generating an invoice, please make sure to send that invoice back to the user.
*   ...

9. Testing the Chatbot

After configuring the AI agent and workflows, test the chatbot to ensure it functions as expected. The example demonstrates requesting an invoice and verifying that it is generated and sent to the user's email.

10. Conclusion

The video concludes by highlighting the power and versatility of NADN for building AI chatbots. It emphasizes the ease of getting started and the ability to extend the chatbot's functionality by adding new tools. The chatbot can be used for various purposes, including lead generation, customer service, and sending invoices or contracts. The presenter also promotes his school community for additional help and resources.

Key Takeaways:

NADN allows for rapid development of AI chatbots with minimal coding.
Chatbots can be easily embedded on websites using a simple JavaScript script.
AI agents can be configured with various memory types and tools to perform specific tasks.
Workflows and webhooks enable seamless integration with third-party services.
System prompts are crucial for guiding the AI agent's behavior and ensuring desired outcomes.