Finally, Google has made AI useful. Notebook LM, an AI research tool that is on fire right now. In this video, I will share some really unique features of Notebook LM that you won't find on other AI research tools, even like perplexity. It's limitations you should be aware of and some of the best ways you should use it over other AI tools for your content research. Let's go.
So what is Notebook LM? It's a beta AI product developed by Google for researching and note-taking purposes. Think of it like your own personalized knowledge base powered by AI.
It allows you to upload different documents and you can start asking questions with the AI that is powered by the latest AI model Gemini 1.5 Pro. It was first officially launched around July, but why is it suddenly getting so popular now? The most impressive unique feature I found is the great support of different sources and multi-modal capability. Like Gemini, it supports lots of file formats like Google Docs, Google Slides, PDF files, images, and now even with YouTube videos, audio files, which is something you won't easily find on other AI like Perplexity. It supports up to 50 sources, making it super useful for researching purposes and to make sure everything generated is not hallucinated.
Because unlike on Perplexity, which is based on searching on the open web or you need to specify the link every time, you're actually first defining the knowledge curve for the AI and so it always refers to the sources you provided. That is super smart. You can just select the source, start asking questions, and you don't need to add or delete every time. And you can just use the building prompt to generate FAQs, study guides, table of contents to organize the database. Or you can even just click on each source and get inspiration on any subtopics it identifies as a starting point.
No, for now, there is no chat history between sections and it will disappear once you start a new session, so always save it as a note if you want to keep it. Also, for some websites, they may block AI access or there is source restriction, like from certain media or publishers, you may not be able to import them. Also, for YouTube videos, only public YouTube videos are supported, unless the videos are not supported. And videos that are recently uploaded less than 72 hours may not be immediately available for import another standout features is definitely its latest audio overview generations basically it summarizes all your sources and turns it into a podcast style discussion between two ai hosts which is super engaging and what makes me so impressed is you will find the voice doesn't sound like a robot superhuman natural making so enjoyable to listen if you're a podcast person like me you will love it you just need to try it out yourself But note that I found this audio generation features, unlike the chatting function, tends to hallucinate information. Because if you listen to it closely, you will find Gemini adds a lot of fluff to make it super conversational.
It may add other irrelevant information to try to make it engaging. So keep this in mind. Currently, there's no way to edit the podcast, change voice, or edit the transcript. But I will show you the workaround. Another thing that I'm super impressed is the citations.
Unlike Proplexly or other AI research tools which it just refer the link as the source and you still need to trace it yourself on a page where it get the information. Notebook LM will refer you exactly where it is from source. You would be surprised how useful this feature is and I'm going to show you how you can use it in a minute. Also, I found the summary generated on Notebook LM is more in-depth than on Germani.
Even on Germani Advice, My guess is because it's still in beta, so less regulation, less quality checking, less filtering, so definitely use it to your advantage. Now here comes the best use cases. The first one is YouTube content research. Since Noble LM supports YouTube video links import, it's really useful for YouTube content research and analyze some of the best performing YouTube videos for the video topic that you're planning. And I found it's even easier than using the YouTube extension on Germani as you can just directly enter the YouTube link.
For example, let's say I'm planning a YouTube video topic about quitting 95. So I will go to YouTube, type in a keyword and then copy the top 10 YouTube videos links because usually they are videos either more engaging or have a high click rate. So there is a lot you can learn from these videos. Now go to notebook LM and create your first notebook by clicking create.
and then under the source import click youtube then paste the 10 youtube videos links one by one unfortunately now there's no bulk import if you want you can double check if the source is successfully imported by checking the transcript it will immediately show here which is really cool then i'll say to summarize the key common themes covered across these videos analyze the commonly used host because this is super useful for successful video And also do you need angles used for each video that you can learn from? And now you can see different common themes, which are good starting points, stress of 95, financial growth potential, and also the hooks. And this is super useful because it gives you ideas on how to grab the viewer's attention and use it in your videos, like sharing personal anecdotes. And also do you need angles which you can take reference? What are the perspective you can take on the topic?
I'm not asking you to copy them. but actually what are some fresh angle you can take on so you are not just copying what others have been saying i love this method because it always works it will immediately pull the transcript so you can always ensure the quality you don't need to copy and paste the transcript every time on cloud or perplexity save you even more time and if you want to get better at using airtools for work and improve your chat gpt skills whether it's for content research creation or any other tasks I highly suggest you check out this bundle from HubSpot that contains 5 essential resources for using ChatGPT at work. I put it in the description for you to download for free.
It has a lot of great resources like a decision flow chart, content refinement checklist, and guidelines on how to use it more effectively for day-to-day work tasks. My personal favorite sections are the content style guideline template and the 100 plus ChatGPT prompts. It gives me directions on how to tune the AI generated content to sound like me. and lots of practical use cases of using AI other than content creation. Especially if you're just getting the hang of AI, using pre-made prompts helps a lot with learning how to use ChatGPT.
You can download this in the description below for free. This bundle was made by HubSpot, which is today's video sponsor. A big shout out to them for providing this free resource. Alright, the next one. Analyze the positioning of your competitors and to improve your own website messaging.
Surprisingly, compared to the perplexity that also has web access. I found the analysis from notebook.lm is more detailed and it's also super handy to just use the website URL import features to do a high-level analysis. Let's say I'm doing competitor research on a topic, past sitting business, then import all the competitor URLs and then ask it to summarize the main message appear across these websites, key service highlighted and the potential target customers.
Turn user in the messaging and then it will give you a detailed summary on the positioning like peace of mind, highlighted surfaces, target audience, their overtone like trustworthiness, compassionate. And what I really like is you can refer to where exactly the copy is. And so you can always double check and nail your own copy. And now import your own website URL. Let's say I'm the owner for this site, Pet Pros Canada.
And I'll say to identify three positioning gaps and suggest three opportunities for these websites to stand out. Make sure you state clearly your website URL in the prompt so it can refer back. Then you will see it summarizes some of the gaps it identifies like lack of mentioning of stock expertise or qualification and propose suggestions like emphasizing the trust factor through the copy. I would say this is much more detailed than what you would get on perplexity or even on German advice.
An important note is never copy your positioning from your competitors. Always identify if any gaps to differentiate yourself and also make sure you always save any responses and notes so you can take reference in the future. The next one is to analyze top random results for content optimization. Again, using its website capture functions, we can use it to analyze the top random pages for target keyword or topic.
And the best thing about it is that you don't even need to copy and paste the text. You just import and do it. Super handy.
For example, I'm planning an article about how to encourage toddlers to talk. And then search on Google, then use this Chrome extension to enable the NS grow and to grab all the URLs for the top random results. There are free plugins, I'll include that in the description.
Then on notebook LM, paste the link one by one. I would say try to analyze at least 20 articles to get a more accurate result by also considering results rendered between position 11 to 20. And like I said, sometimes not every web URL can be imported. So either you copy and paste the text here or you just abandon this source. Like in this case, we'll just remove that and add another one until we get 20. Once you import, always double check if they are successfully imported.
And now, ask it to analyze these top results and summarize the main topics covered which we can also use in the blog. Also suggest a list of systematic keywords to increase the content relevance around the primary target keyword. And then you will have really detailed main topic ideas that inspire your content planning for your article. And what I really like is that you can click the citation and immediately refer back to where exactly this point is made super useful features that you won't find on perplexity what i always suggest is don't just copy these ideas never just copy but how you can expand on these points or ask ai to give you some other talking points that aren't covered yet so making the content more unique and then for semantic keywords such as speech delay language enrichment activities these keywords will be super useful to increase content relevance.
And when you want to rank your website on AI power search, if you want, you can even ask it to give it more keywords to further expand the pool. The next one, turning a blog article into a podcast. The audio overview generation features is great, but how could you really use it in your own podcast?
Because obviously you won't use the same audio overview in your podcast with the two AI hosts, right? And so I'm going to show you the workaround. Let's say I have this article about free keyword research tools and I want to turn it into a podcast format. Then on Notebook LM, add it as a website source.
Then click generate on the audio overview. For the sake of demo, I'm not going to play it. Download the audio file and we're gonna re-upload it again as another source. The goal here is to get the transcript.
So you can just click it and you will see the transcript is loaded. Copy it and we're going to edit it on clock. Now, on cloud or any AI tools that you like, perhaps chat GPT, ask it to turn the script using a first-person point of view as if I am the host, Grace, and remind it to keep the original version as much as possible.
This is super important because we want to keep the conversational style just like in the original version. So you can see this is great. I love the intro and some of the sections are really good actually, although I can see some editing is still needed. That's already a good starting point.
I always suggest fine-tuning it further using cloud until you're satisfied. So once you have the final script, you can then record it yourself or if you want, you can even use some text-to-speech AI tools like Speechify Descript to clone your own voice and turn it into speech instantly. Though I always suggest to use your own voice whenever possible so that at least to keep some human elements to make it authentic. To achieve the best results, always use your original content and I don't suggest you generate the podcast script using any purely AI generated content because it is super important for building a real audience and not just an audio file so you are only using Noble LM to summarize the original content insights and fast track the script generation process the next one analyze LinkedIn comments for audience research you know oftentimes for the content research process the goal is usually hidden in the comments but Plexly may be able to analyze the post but not comments so this is a big plus when using noble lm first of all identify the top linkedin influencers in your niche because usually they have a big following which may be your target audience and so you can learn a lot from the post comments let's say i'm doing kind of research in leadership and for sure simon scenic is one of the linkedin top force in this niche then analyze what are the posts that are obviously doing better than others like this one having over 10 000 engagements And then you can copy the link for this post and paste it back on notebook lm.
Note that unlike other paste scraping tools that scrape all the comments, this method is only able to analyze the comments that are visible on the site. And so that's nine comments per post and that's the limitation. So for the same LinkedIn profile, I recommend analyzing at least 10 posts or more. And now we have all the 10 post links.
Then ask it to analyze the comments and to identify the main points that resonate the most with the audience. Highlights if any disagreement, their desires, and to suggest 5 more LinkedIn post content ideas based on these analysis. Alright, I think this is really detailed.
It even points out who made this comment and you can click and check it. This is really helpful. And they are all good. insights like vulnerability, authenticity, over-perfection, and the audience wants more practical guidance on how to be vulnerable as a leader, then finally come up with ideas. This is already perfect.
Again, the goal is not to analyze the LinkedIn post itself, but through the comments, we understand more what the target audience wants. So think about any well-known thought leaders in your niche and apply this for your content research process. To achieve the best results, I recommend finding at least 3 top voices and for each of them, analyze at least 10 posts so you can make sure you get more data for more accurate results. And you can apply the same to analyze Reddit comments. Just find some top Reddit posts around your topic and do the same analysis on notebook LLM.
Alright, don't just take my word for it. Definitely try all these use cases yourself. And before you go, if you constantly need to do content research, also watch this video about using perplexity AI so you can better combine both of them to level up your whole research process. I'll see you next time.