Built an AI system that reorganized everything so I could find what I need

AI Mastery Lesson Summary System: From Notebook LM to Obsidian

Project Owner: Tracie Fujikane

My Struggles Transformed into a Learning Revolution

I struggled in school with a brain that worked differently. Back then, they didn’t diagnose learning disabilities; instead, they called you stupid, lazy or not trying hard enough. 

Those early challenges shaped the way I think today and sparked my passion for creating tools that make learning more accessible. AI and automation have become game changers for anyone who struggles to fit into a traditional learning environment. This project AI Mastery Lesson Summary System – is the culmination of that journey. It uses technology as an “external Brain” to help others understand, retain, and apply complex concepts with ease.  

If this system can bring clarity to even a few people who feel overwhelmed by dense material, it will have achieved its purpose – and made a meaningful difference in their learning journey!

Project Collaborators: This project was a collaborative effort, supported by the dedication and contributions of:

- Marcela Herrera

- Cosmina Balu

- Denï D’Ambrogio

Project Description:

The foundation of this project began with the goal of automating the creation of detailed lesson summary pages in Notion for each session of the Mindvalley AI Mastery Advanced Course. The initial aim was to leverage AI-driven automation to streamline the learning process by organizing and enhancing materials into structured, indexed Notion pages. These pages featured key components such as:

·       Key Points Summary: A concise overview of the most important takeaways.

·       Comprehensive Overview: A detailed breakdown of lesson content, including core concepts and insights.

·       Examples and Case Studies: Real-world applications to deepen understanding.

·       Practical Applications: Tailored suggestions for applying lesson content.

·       Action Items: Clear, actionable steps to implement the knowledge gained.

·       List of Resources/Tools: Curated additional resources mentioned or recommended.

This original concept evolved through multiple iterations, transforming into a robust system that enhanced learning accessibility, efficiency, and personalization. By incorporating tools like Notion, Notebook LM, and Obsidian, the project expanded its functionality to include advanced querying, timestamp-linked references, and offline repositories, showcasing the potential of AI to revolutionize the learning experience.

Imagine having a personal assistant that turns hours of course content into a crystal-clear, easy-to-navigate guide. That’s exactly what the AI Mastery Lesson Summary System does!

This project has been and continues to be a journey of innovation and iteration, progressing from one concept and transforming throughout our learning during the course to incorporate new concepts and tools. At each step, we overcame challenges and expanded our capabilities, moving us closer to a solution that enhances learning accessibility, efficiency, and personalization.

By leveraging tools like Notion, Notebook LM, and Obsidian, the system transformed raw transcripts and course materials into structured, actionable guides.

Each iteration added new functionalities and options, such as the ability to query our materials, produce timestamp-linked references, and create offline repositories, demonstrating the potential of AI to revolutionize learning.

Tools Used:

- Notebook LM (Powered by Gemini 2.0): Core AI platform for summarizing, analyzing, and connecting materials.

- Notion: Organized structured lesson summaries and Loom guides.

- Obsidian: Provided offline access and real-time AI capabilities.

- Google Drive: Centralized file storage for seamless collaboration.

- OpenAI API: Removed timestamps to reduce the token size of transcripts and create the structured summaries.

- Make: Automated mapping of summaries to Notion.

- Guidde: Created AI-powered instructional guides to explain the workflow.

- See the following link for the AI voiced-over video and step by step guide: Guidde Playbook.

- FreeConvert: Streamlined DOCX to TXT conversion for AI processing.

- PDF.co: Converted PDF so that OpenAI was able to process the transcripts.

In Progress:

- n8n Automation: Developing workflows to bring transcripts into Obsidian.

- FreeConvert API: Exploring automation to enhance file conversions for seamless processing.

Iterations:

1st Iteration: Notion

- Goal: Organize and structure indexed course content, using automation, for easy access and navigation.

- Approach:

- Integrated structured lesson summaries into Notion.

- Used Make to automate retrieving transcripts, removing timestamps, summarizing, and populating Notion.

- Challenges:

- Token limits.

- File formats need to be converted.

- Mapping structured data points into a comprehensive document in Notion.

- Outcome: Created a user-friendly index and summary pages for lessons and Loom topics.

-

2nd Iteration: Notebook LM Plus

- Goal: Enhance learning with AI-driven capabilities for querying, summarizing, chatting with notebooks, and creating actionable insights and audio podcasts.

- Approach:

- Leveraged Notebook LM to process 181 reference sources.

- Created overview documents as sources to give more context to the material.

- Generated detailed lesson guides using tailored prompts.

- Introduced audio summaries and timestamp-linked references.

- Used the Beta feature, Interactive Mode, to verbally interact with the podcast moderators to direct the discussion in real time. (Current limitation: Interactions are not recorded or documented.)

- Challenges:

- Requires a paid account to accommodate more than 50 sources.

- No API available for automation or integration.

- Audio podcasts must be manually generated one at a time and deleted to create the next podcast.

- Outcome: A dynamic experience and engagement with materials through a more interactive and personalized experience.

3rd Iteration: Obsidian - Still Iterating

- Goal: Provide an offline, secure repository for personal knowledge management.

- Approach:

- Imported all course materials into Obsidian.

- Organized data to complement online solutions.

- Incorporated community plug ins to leverage both local and online LLM models to allow querying of the course material, as well as expanding on the topics and scenarios through including real-time data and up to date resources.

- Outcome: Created a centralized, locally hosted, searchable knowledge base for long-term use, while expanding the scope of reference to be beneficial beyond the literal course material; and in the context of other notes unrelated to the AI Mastery course.

- The Next Iteration: Complete an n8n automation to bring transcripts into the vault, use the FreeConvert API to convert docx to txt, and use an Obsidian community plugin to create a summary template that will run a script when a new transcript is uploaded to the transcript folder.

Methodology:

The process behind the AI Mastery Lesson Summary System was methodical and iterative, designed to address the evolving needs of learners. Here are the key phases:

1.     Data Collection and Cleaning: Transcripts, Loom videos, and supplemental materials from the AI Mastery course were sourced and organized in Google Drive. Data cleaning was a critical step—manually scrubbing the data to ensure consistency and accuracy required significant time and effort but was essential to ensure accuracy in our results. Finding words that were incorrect, but not identified as misspelled, such as “irritable” instead of Airtable, “flow eyes” instead of Flowise, are just a couple of examples. Clean data is king, and this foundation supported all subsequent automation and AI analysis.

2.     Preprocessing: FreeConvert was used to convert DOCX files into TXT format for seamless integration with AI tools.

3.     Automation Development with Make for Notion: Initially, a simple Make workflow was created, but it didn’t produce the desired results. The introduction of the JSON module, along with additional steps, smoothed out the process. This iterative improvement transformed the automation into a robust system capable of handling complex workflows.

1. Notebook LM Integration: AI-powered tools generated summaries, interactive guides, and timestamp-linked references for personalized learning experiences. A standardized prompt template was used to create static notes that summarized each lesson and each loom video.

2. Obsidian Implementation: Materials were integrated into Obsidian for offline use, with plugins to enable advanced querying and contextual learning.

3. Feedback and Iteration: User feedback was used to refine tools, prompts, and integrations across platforms.

Achievements:

- Iteratively refined the system to address user needs and platform constraints.

- Delivered a flexible, multi-platform solution, catering to diverse learning preferences.

- Empowered learners with innovative tools like timestamp-linked references and audio guides.

- Enhanced accessibility through features supporting various learning modalities.

Learnings:

- The Power of Iteration: This project demonstrated the importance of continuous improvement. Each iteration—Notion, Notebook LM, and Obsidian—addressed specific challenges and introduced new features, resulting in a multi-layered solution that caters to different learning styles and needs.

- Balancing Automation and Manual Effort: While automation reduced manual work, limitations such as token constraints, lack of APIs, and manual podcast generation highlighted areas where human intervention was necessary. These challenges informed our understanding of where technology currently falls short and where future innovations are needed.

- Personalized Learning is Transformational: By integrating interactive tools like Notebook LM’s chat feature and Obsidian’s plug-ins, we saw firsthand how personalized approaches, like tailored queries and contextual expansion, can deepen engagement and understanding.

- Accessibility and Flexibility are Key: Offering content across multiple platforms (Notion, Notebook LM, Obsidian) ensured the system met diverse needs, from online collaboration to offline knowledge management. This reinforced the idea that learning tools should adapt to the user, not the other way around.

- Collaboration Amplifies Success: The contributions of team members and the feedback from course participants were invaluable. This collaborative effort highlighted the strength of combining diverse perspectives to solve complex problems.

- AI's Role in Lifelong Learning: Leveraging AI to query, summarize, and expand on course materials showed its potential as an “external brain,” enabling learners to overcome limitations and maximize retention and recall.

Real Voices: 

Here’s how people are benefiting from the “AI Mastery Notebook LM”

Krista Korinke:

"I like being able to ask, via natural language chat, where to find something I remember from the AI Mastery course, without having to figure out which lesson it was in. For example, I asked which lesson had the Mars chatbot, and it told me Lesson 15. Being able to locate information or instructions easily is one of the best use cases."

Steve Finegan:

"I asked a question about a specific use case I was working on. For example, if I want to know how to create an automation for generating images or mapping data in an Airtable base to a Google Doc, I can ask it. Notebook LM tells me which videos or other materials have the information I need."

Jay Bilakhia:

"I’ve found that it can be used in various ways depending on the need. I plan to utilize it for organizing and tracking goals while maintaining daily reflections."

Celia Bohle:

"An excellent retrieval system—it efficiently surfaces relevant content from the vast course material. The integration of Loom transcripts, which I didn’t have before, is a great bonus! The Studio section is fantastic for content summaries."

Denï D’Ambrogio:

"Amazing tool! I have ADHD, and I can be a mess, but this project reorganized everything in a way that makes it easy for me to find what I need. It’s incredible - a bible of the wonderful AI Mastery experience. I’m forever thankful for this gift."

Robert Granoff:

“I can’t imagine a better tool than Notebook LM to organize the enormous scope of information that was covered in the 4-month class.  Not only are the class transcripts included but also office hour transcripts and loom transcripts!  Nothing has been left out.  It’s a perfect showcase with which to revisit specifics and also overall information.”

Gwen Mitchell:

“I’m using Notebook LM Plus to capture all the 3 Mastery programs I’m in so it will be easy to quickly reference materials or call up a technique … again thanks for turning me on to this tool!!!

Lana :

“I must say, I’m truly impressed with the tool! The speed at which you can go through all the lessons is amazing. The ability to upload various sources and instantly generate summaries, podcasts, and outlines makes it incredibly practical and user-friendly.”

URLs: Guidde - https://app.guidde.com/share/playbooks/ixQSUGzzGXQSxt8UZSjsdW?origin=zNZulq0lQTTidMVEnPFD5LyiWfc2

Notebook LM - https://notebooklm.google.com/notebook/176950bc-7475-408b-83c3-1321b4aa2446

*To be granted access to the Notebook, provide a GMAIL email address that can be added.