This is insane and I wish I would have read something like this earlier!
I guess it is visible based on my profile that I'm more or less a health nerd and I dig deep into studies, I love reading stuff on PubMed. I spend a lot of my time there and obviously I'm trying to get better at using certain AI tools for research, but recently I noticed that I have been super inefficient because I had plenty of Perplexity queries open doing research, finding papers, analysing them for me and a fraction is not much but even that did not stick in my brain upon reading the summaries, looking through the tables and figures.
It was so inefficient and I was thinking I wish there was something that summarizes and discusses the topics like a video or a podcast or something like that. And I never even heard of NotebookLM before.
So, I'll give this method a shot, although I haven't really been using Gemini for deep research just yet, only Perplexity. But I think this could help me a lot to dive deeper into the topic truly interested in and probably wasted a lot of time on learning deeply but with no signs of efficiency whatsoever.
This is a brilliant use case. NotebookLM is my favorite AI tool. I use it to extract frameworks and have conversations with PDF books. I generally create agents (Gemini gems) and use that for research. I love the idea of capturing that research and bringing it back into a notebook. I’m going to give that a try. 🙏
When I create Gemini Gems, I model them on frameworks extracted from NotebookLM. For example, I’ll build a NotebookLM around Marty Cagan, extract his mental models, install those into a Gemini Gem, and then use the Gem for conversation and research. What I don’t do is feed the research generated by the Gemini Gem back into NotebookLM for synthesis or later consumption.
Love this! I do something similar with ChatGPT, Perplexity, Claude and Notion. Research. Document. Analyze. Summarize. Simplify. Rinse and Repeat.
I do listen to my summaries via voice on ChatGPT but Gemini and NotebookLM are going to be placed in rotation.
Solid article!
Wow, this is wild. Guess I need to get NotebookLM now 😂
It is free!
Sounds awesome! I'm going to try this. Thanks!
This is insane and I wish I would have read something like this earlier!
I guess it is visible based on my profile that I'm more or less a health nerd and I dig deep into studies, I love reading stuff on PubMed. I spend a lot of my time there and obviously I'm trying to get better at using certain AI tools for research, but recently I noticed that I have been super inefficient because I had plenty of Perplexity queries open doing research, finding papers, analysing them for me and a fraction is not much but even that did not stick in my brain upon reading the summaries, looking through the tables and figures.
It was so inefficient and I was thinking I wish there was something that summarizes and discusses the topics like a video or a podcast or something like that. And I never even heard of NotebookLM before.
So, I'll give this method a shot, although I haven't really been using Gemini for deep research just yet, only Perplexity. But I think this could help me a lot to dive deeper into the topic truly interested in and probably wasted a lot of time on learning deeply but with no signs of efficiency whatsoever.
I am.curious to get your feedback on how it worked.
Genius
This is a brilliant use case. NotebookLM is my favorite AI tool. I use it to extract frameworks and have conversations with PDF books. I generally create agents (Gemini gems) and use that for research. I love the idea of capturing that research and bringing it back into a notebook. I’m going to give that a try. 🙏
So you are using the two tools but not bridging them yet?
When I create Gemini Gems, I model them on frameworks extracted from NotebookLM. For example, I’ll build a NotebookLM around Marty Cagan, extract his mental models, install those into a Gemini Gem, and then use the Gem for conversation and research. What I don’t do is feed the research generated by the Gemini Gem back into NotebookLM for synthesis or later consumption.
You were probably just getting overloaded with information. That's one downside to having all these tools with near infinite information.
Let me know how it goes.
Thanks for the feedback
Oh that's incredible.
I have yet to build a persona based on the knowledge of a real person. I may try this.
I always just assumed you are asking Claude to make a profile based off its training data or open source material it will never have proper context.
What material do you use to feed your notebookLM? The person's own writings?