## When to use NotebookLM [(00:00:00)](https://www.youtube.com/watch?v=EOmgC3-hznM&t=0s) - [[NotebookLM]] is suitable for use cases that match three specific criteria: having a low tolerance for hallucination, working with scattered information across different locations and formats, and needing a quick and reliable way to transform fragmented information into a cohesive output [(00:00:26)](https://www.youtube.com/watch?v=EOmgC3-hznM&t=26s). - The first criterion for using NotebookLM is having a very low tolerance for hallucination, which is a key consideration for users who require accurate information [(00:00:08)](https://www.youtube.com/watch?v=EOmgC3-hznM&t=8s). - The second criterion involves working with information scattered across different locations, formats, and mediums, such as documents, slides, text, video, and audio [(00:00:16)](https://www.youtube.com/watch?v=EOmgC3-hznM&t=16s). - The third criterion is the need for a quick and reliable way to transform fragmented information into a cohesive and meaningful output [(00:00:22)](https://www.youtube.com/watch?v=EOmgC3-hznM&t=22s). - NotebookLM is expected to perform better than other AI tools when these three criteria are met [(00:00:05)](https://www.youtube.com/watch?v=EOmgC3-hznM&t=5s). ## NotebookLM Basics [(00:00:36)](https://www.youtube.com/watch?v=EOmgC3-hznM&t=36s) - When creating a new notebook, it can be overwhelming for users to see all the options available, so it's recommended to press escape, click the logo in the top left corner to go to the Notebook LM homepage, and bookmark it for easy reference [(00:00:37)](https://www.youtube.com/watch?v=EOmgC3-hznM&t=37s). - To create a new notebook, press escape, name the notebook, and click the plus icon to start uploading sources, which can include files and [[YouTube]] videos [(00:00:58)](https://www.youtube.com/watch?v=EOmgC3-hznM&t=58s). - Sources can be uploaded and processed, and then individual sources can be clicked into to view a concise summary produced by Notebook LM [(00:01:39)](https://www.youtube.com/watch?v=EOmgC3-hznM&t=99s). - Under the Source guide, Notebook LM can expand on a topic based on all selected sources, and everything in the chat interface takes into account all selected sources [(00:01:49)](https://www.youtube.com/watch?v=EOmgC3-hznM&t=109s). - If a source needs to be ignored, it must be deselected from the Source list first [(00:02:19)](https://www.youtube.com/watch?v=EOmgC3-hznM&t=139s). - The Notebook guide feature provides a quick start guide for beginners, including a summary of all sources, pre-created templates, and suggested questions to help users get started [(00:02:35)](https://www.youtube.com/watch?v=EOmgC3-hznM&t=155s). - The Notebook LM homepage can be customized by changing to list view, sorting by title, and creating a new notebook [(00:00:49)](https://www.youtube.com/watch?v=EOmgC3-hznM&t=49s). - Sources can be added from various formats, including PDF files, such as "[[Tools of Titans]]" by [[Tim Ferriss | Tim Ferris]], and [[YouTube]] videos on specific topics, such as uric acid and fasting by [[Andrew Huberman]] [(00:01:13)](https://www.youtube.com/watch?v=EOmgC3-hznM&t=73s). ## Step-by-step Workflow [(00:02:56)](https://www.youtube.com/watch?v=EOmgC3-hznM&t=176s) - A user interacts with a Health reports notebook, asking for the top 10 Health Trends based on their last three reports, and [[NotebookLM]] generates a list of 10 observations in a few seconds [(00:03:04)](https://www.youtube.com/watch?v=EOmgC3-hznM&t=184s). - The user is informed that they have had elevated uric acid levels over the last 3 years, which is true due to their diet, and they can ask what they can do to lower their uric acid levels [(00:03:16)](https://www.youtube.com/watch?v=EOmgC3-hznM&t=196s). - NotebookLM provides three suggestions to lower uric acid levels, and the user can click on an inline citation to view the video transcript that supports this information [(00:03:24)](https://www.youtube.com/watch?v=EOmgC3-hznM&t=204s). - It is essential to click "Save to note" to refer back to good outputs, as NotebookLM is not trained on uploaded data, including conversations, and outputs disappear when the notebook is reloaded [(00:03:50)](https://www.youtube.com/watch?v=EOmgC3-hznM&t=230s). - The user asks if their health reports suggest they shouldn't fast for 36 hours every week, and NotebookLM shares context, stating that their blood work is normal and they can continue fasting with no issues [(00:04:17)](https://www.youtube.com/watch?v=EOmgC3-hznM&t=257s). - NotebookLM can quickly and efficiently reference multiple sources, connect relevant dots, and produce a good enough output in significantly less time compared to manual analysis [(00:04:43)](https://www.youtube.com/watch?v=EOmgC3-hznM&t=283s). - When prompted with a question about "inerting D Lo weeks" into their workout routine, [[NotebookLM]] states that none of the sources say anything about it, but mentions related topics [(00:04:56)](https://www.youtube.com/watch?v=EOmgC3-hznM&t=296s). - This illustrates that NotebookLM is fine-tuned to hallucinate less but is also less creative, whereas [[Gemini (chatbot) | Google Gemini]] is optimized for speed and creativity [(00:05:17)](https://www.youtube.com/watch?v=EOmgC3-hznM&t=317s). - Users can manually add a note, select multiple notes, or select all and convert them to a standalone source, and then copy and paste the text to use elsewhere [(00:05:32)](https://www.youtube.com/watch?v=EOmgC3-hznM&t=332s). ## Focused Knowledge Retrieval [(00:06:00)](https://www.youtube.com/watch?v=EOmgC3-hznM&t=360s) - Focused knowledge retrieval is a feature that allows users to ask specific questions and retrieve relevant information from their sources, such as user manuals, tax codes, and interview notes [(00:06:01)](https://www.youtube.com/watch?v=EOmgC3-hznM&t=361s). - To replicate this feature, users can [[Google]] the product they have followed by "user manual" and "PDF" to find relevant information, or add the website directly as a source [(00:06:26)](https://www.youtube.com/watch?v=EOmgC3-hznM&t=386s). - Some websites may block [[NotebookLM]] from adding them as sources, but users can get around this by adding another source and copying and pasting the text from the website [(00:06:49)](https://www.youtube.com/watch?v=EOmgC3-hznM&t=409s). - Users can create notebooks for specific topics, such as equipment manuals, tax and accounting, or recruiting, and add relevant sources to ask personalized questions [(00:06:04)](https://www.youtube.com/watch?v=EOmgC3-hznM&t=364s). - For example, a tax and accounting notebook can include documentations like tax codes from the government and audit reports from accounts, allowing users to ask questions like "what are my tax obligations last year" [(00:07:00)](https://www.youtube.com/watch?v=EOmgC3-hznM&t=420s). - A recruiting notebook can include sources like guidelines from HR, performance rubrics, question banks, candidate resumes, and interview notes, allowing users to ask questions like "what are the key achievements and relevant skills of this candidate" [(00:07:27)](https://www.youtube.com/watch?v=EOmgC3-hznM&t=447s). - When using this feature, it's essential to only select the documents from the candidate being interviewed to avoid incorporating information from other candidates [(00:08:00)](https://www.youtube.com/watch?v=EOmgC3-hznM&t=480s). ## Project Context Engine [(00:08:09)](https://www.youtube.com/watch?v=EOmgC3-hznM&t=489s) - The Project Context Engine use case involves having a notebook for each project at work, where meeting notes, project plans, and documents from similar projects are added as sources [(00:08:10)](https://www.youtube.com/watch?v=EOmgC3-hznM&t=490s). - Project and program managers can benefit greatly from [[NotebookLM]] as it helps them work with scattered information and synthesize it into an easy-to-digest format [(00:08:21)](https://www.youtube.com/watch?v=EOmgC3-hznM&t=501s). - NotebookLM can be used to create high-level briefing documents for senior leaders, campaign timelines, and FAQ documents for colleagues unfamiliar with the project [(00:08:43)](https://www.youtube.com/watch?v=EOmgC3-hznM&t=523s). - Uploading meeting transcripts from [[Zoom Video Communications | Zoom]] or [[Google]] Meet can unlock highly accurate answers to questions about outstanding tasks or meeting recaps [(00:09:01)](https://www.youtube.com/watch?v=EOmgC3-hznM&t=541s). - NotebookLM can identify learnings and strategies from previous projects by analyzing shared recap documents, which can be incorporated into future campaigns [(00:09:17)](https://www.youtube.com/watch?v=EOmgC3-hznM&t=557s). - To get started with NotebookLM, it is recommended to upload related files and try the suggested questions, which can be surprisingly helpful [(00:09:26)](https://www.youtube.com/watch?v=EOmgC3-hznM&t=566s). - Google Workspace users can join a Weekly Newsletter to receive actionable tips every week, with a link provided below [(00:09:38)](https://www.youtube.com/watch?v=EOmgC3-hznM&t=578s). ## Targeted Insights Studio [(00:09:46)](https://www.youtube.com/watch?v=EOmgC3-hznM&t=586s) - A solution was created to stay current with AI developments by making an earnings analysis notebook with earnings reports from tech companies and articles from tech analysts compiled in a [[Google Docs]] format, allowing for targeted questions to be asked using Notebook LM [(00:09:47)](https://www.youtube.com/watch?v=EOmgC3-hznM&t=587s). - Notebook LM can be used to ask very targeted questions, such as Google's monetization strategy with regards to AI, and receive a structured list in response [(00:10:16)](https://www.youtube.com/watch?v=EOmgC3-hznM&t=616s). - Broader questions can also be asked, such as how AI strategies differ for [[Google]], [[Facebook | Meta]], [[Amazon (company) | Amazon]], and [[Apple Inc. | Apple]], and Notebook LM will provide a one-sentence summary before expanding on the respective strategies [(00:10:23)](https://www.youtube.com/watch?v=EOmgC3-hznM&t=623s). - Notebook LM can generate a personalized podcast episode by clicking "customize audio overview" and providing specific instructions, such as focusing on how earnings from one company affect its competitors and assuming the listener has zero technical background [(00:10:45)](https://www.youtube.com/watch?v=EOmgC3-hznM&t=645s). - The ability to generate a personalized podcast episode is an example of how AI can be used to improve existing products rather than replacing them, which is a strategy that Meta seems to understand [(00:11:02)](https://www.youtube.com/watch?v=EOmgC3-hznM&t=662s). - When adding a Google Doc or slide as a source, it is possible to click and resync the file after changes are made to ensure that the most up-to-date information is being used [(00:11:20)](https://www.youtube.com/watch?v=EOmgC3-hznM&t=680s). ## Beyond the Basics [(00:11:33)](https://www.youtube.com/watch?v=EOmgC3-hznM&t=693s) - Notebook LM rarely hallucinates, but it's not optimized for creativity, so it's often necessary to use other tools like Gemini or Claw to produce the final deliverable [(00:11:34)](https://www.youtube.com/watch?v=EOmgC3-hznM&t=694s). - The amount of information Notebook LM can absorb is massive, around 25 million words per notebook, compared to 500,000 words for Gemini, 100,000 for Claw, and 64,000 for Chach BT [(00:11:47)](https://www.youtube.com/watch?v=EOmgC3-hznM&t=707s). - Although there's a limit of 20 sources per notebook, it's possible to combine multiple documents into one file to work around this limitation [(00:11:58)](https://www.youtube.com/watch?v=EOmgC3-hznM&t=718s). - The quality of the sources used with Notebook LM is extremely important, and using articles from well-established publications will generally produce better outputs than low-quality blog posts [(00:12:09)](https://www.youtube.com/watch?v=EOmgC3-hznM&t=729s). ## Sources - [website](https://www.youtube.com/watch?v=EOmgC3-hznM?si=cH7IYrPOby1_yJ7H)