VIDEO
HIGHLIGHT
Log InSign up
I Built an AI Alex Hormozi I Can Talk To On Demand (No Code RAG Agent in n8n)
SummaryTranscriptChat

I Built an AI Alex Hormozi I Can Talk To On Demand (No Code RAG Agent in n8n)

How to Create a Chatbot Using YouTube Video Transcripts

Introduction to the Tool

  • The speaker introduces a tool that allows users to interact with their favorite YouTubers by asking questions and receiving answers based on video content.
  • This tool utilizes a RAG (Retrieval-Augmented Generation) database, which scrapes video transcripts from any YouTube channel, enabling chatbot functionality that mimics the YouTuber's knowledge.

Functionality of the Chatbot

  • Demonstration of the chatbot in action, showcasing how it retrieves information in the voice of Alex Hormozi based on his collective knowledge.
  • Discussion on pricing strategies, emphasizing the importance of value-to-price delta for creating attractive offers.

Benefits for Content Creators

  • The system can help YouTubers engage with their audience more effectively without needing constant presence, enhancing community interaction.
  • The speaker invites viewers interested in similar systems for their businesses to schedule discovery calls for personalized solutions.

Overview of Workflow Steps

  • The workflow consists of three main parts: gathering video URLs, obtaining transcripts, and building out the chatbot.

Step 1: Gathering Video URLs

  • Initiates with a manual trigger using self-hosted NAN or an HTTP request via Apify to scrape data from YouTube channels.
  • Apify is highlighted as a powerful scraping tool; its free plan is sufficient for small channels.

Step 2: Scraping Video Data

  • Two scrapers are used: one costs $0.50 per thousand videos and focuses only on long-form content while excluding shorts.

Step 3: Processing Data into Google Sheets

  • After scraping, results are processed through nodes that check execution status and retrieve items from datasets before splitting them into individual entries.
  • Final output includes a Google Sheet containing all video titles and URLs from the selected YouTuber’s channel.

Next Steps in Workflow Execution

Transcribing and Storing YouTube Videos Efficiently

Workflow for Transcription Management

  • The process begins with ensuring that previously transcribed videos are not re-transcribed, allowing for a seamless continuation of workflows if interrupted.
  • A loop is established to handle each transcript individually, preventing data overload in N8N, Ampify, and Superbase systems.

Utilizing Apify for Transcript Scraping

  • The YouTube Transcript Ninja actor is employed at a cost-effective rate ($10 for 1,000 results), which is suitable given the volume of videos (approximately 450).
  • Configuration includes setting the URL from the loop over items node and adjusting settings such as timestamps inclusion and language defaults.

Integrating with Superbase

  • After confirming completion of transcript retrieval, the next step involves sending data to the Superbase vector store.
  • A new project must be created in Superbase (e.g., "Hormosbot") along with setting up a rag table using provided SQL scripts from documentation.

Setting Up Credentials and Data Insertion

  • Project URL and service role secret are required to establish credentials in N8N; testing ensures successful connection to Superbase.
  • The operation within the vector store node focuses on inserting documents into the "documents" table while maintaining default batch sizes.

Finalizing Data Processing Steps

  • An embeddings node is connected using OpenAI embeddings; model selection defaults to text embedding 3 small for general use cases.
  • A data loader is configured to load specific JSON data including video titles and transcripts obtained from Apify nodes.

Completing Video Processing Loop

  • Each processed video updates Google Sheets by marking it as done based on matching video URLs from Apify nodes.
  • This looping continues until all videos are processed and stored in the Superbase vector store.

Creating an AI Chatbot Using Transcripts

Chatbot Configuration Overview

  • The chatbot setup includes a chat trigger linked to an AI agent utilizing the Superbase vector store as its tool for retrieving documents.

Prompt Design for Effective Responses

  • A detailed prompt instructing the AI to respond as Alex Hormosi emphasizes leveraging knowledge from transcripts while ensuring responses maintain relevance through extrapolation.

Importance of Extrapolation in Responses

How to Enhance Your Workflow with ChatGPT

Leveraging ChatGPT for Powerful Workflows

  • The workflow discussed utilizes ChatGPT to provide answers, enhancing the overall power of the application. A template for this workflow is available upon request.

Expanding Data Sources for Enhanced Knowledge

  • Currently, only transcripts from Alex Hormozy's videos are being used; however, there is potential to include all blog posts, tweets, and Instagram posts from him into the RAG (Retrieval-Augmented Generation) database.

Automation for Continuous Updates

Video description

Book a discovery call: https://calendly.com/shabbirnoor/15min 💬 Want the template? Comment! 📢 Subscribe for more no-code AI automation tutorials like this one. What if you could ask Alex Hormozi anything — and get an answer in seconds? In this video, I’ll show you how I built a personal AI Hormozi agent using Apify, OpenAI, and n8n — completely no-code. The agent scrapes Hormozi’s YouTube videos, pulls transcripts, stores them in a RAG (Retrieval-Augmented Generation) system, and lets me chat with the content as if I’m talking directly to him. 🚀 Here’s what the agent does: • Scrapes YouTube videos using Apify • Extracts video transcripts with Apify + automation • Loads all content into a RAG-enabled vector store • Lets you chat with the content via OpenAI in n8n This framework works with any YouTuber. You can build your own personal AI mentor — or create chatbots powered by real voices and content.