Most firms are not comfortable using NotebookLM due to data sovereignty issues otherwise it’s the best tool where you can just upload all company knowledge and achieve what you want without having to pay for enterprise licenses
Thanks Joel, you're one of the few people who I trust absolutely for their expertise about how to use AI tools effectively, and NotebookLM is becoming my go-to tool for actual research. That statistic about the amount of money that hallucinations are costing companies is absolutely wild. Surely it can't be long before more people pick up on the true potential of notebookLM in business and beyond.
NotebookLM is incredible, but it needs time to cook. I load everything first (PDFs, docs, CSVs, images, notes), do a quick citation sanity check, then let it sit 24 hours in the smoker before I ask for Studio outputs. Sometimes 36–48 if the corpus is big 📐
Thank you Joel for showing the precision side of NotebookLM and how it does not hallucinate by design! The more I feel I have used this tool, the more updates it gets every month! 😁
Thank you, you gave us a very clear strategy on how to implement Notebook LM. It will be very helpful to my personal practice as a teacher and researcher.
Hi Joel. Yes, I have. And I am already thinking about ideas to implement it. I just helped a masters student that I advise on creating a notebook to upload all the references and videos he will use in his literature review. I hope Notebook LM is able to read the books in pdf format.
Notebook LLM is one of the most powerful AI tools on the market for learning and understanding in my opinion. Thanks for giving us even more tips and tricks for utilizing it to the max!
Poppy AI runs many different ones; it runs some Anthropic, some OpenAI. It’s similar to Notebook LM, but honestly wouldn't surprise me if at some point in the future they are acquired by Google or one of these bigger engines.
This is good, especially the idea that leaders must treat NotebookLM as a research system not a chatbot. That distinction matters when decisions are on the line.
From my experience in delivery and tech leadership, tools that force grounded context and verifiable outputs change how teams converge on strategy - whether it’s onboarding, competitive insights, or product roadmaps.
NotebookLM’s source-grounded approach (where every insight ties back to actual docs) is exactly the antidote to “fluffy confidence” we see with general LLMs.
It’s not just about avoiding hallucinations - it’s about trustworthy operational workflows that scale across teams and reduce risk.
Thanks, Joel, for this post. A few takeaways: I haven’t used persona customization in NLM yet, so that was a great idea, and I clearly need to dig more into Studio, I’ve been experimenting with charts but not the more creative parts. Just yesterday I uploaded a 4‑hour YouTube video I didn’t have time to watch, asked for a few key points, and it delivered exactly what I needed. Now I can build on that and add more content to complete my case.
Most firms are not comfortable using NotebookLM due to data sovereignty issues otherwise it’s the best tool where you can just upload all company knowledge and achieve what you want without having to pay for enterprise licenses
But data stays private with notebooklm, or as private as Google Drive and other Google products
Thanks Joel, you're one of the few people who I trust absolutely for their expertise about how to use AI tools effectively, and NotebookLM is becoming my go-to tool for actual research. That statistic about the amount of money that hallucinations are costing companies is absolutely wild. Surely it can't be long before more people pick up on the true potential of notebookLM in business and beyond.
I appreciate that, Sam! To me how much you care about hallucinations really comes down to what's the value of your credibility.
Absolutely this. I loved the note you had about this as well. I really do try to validate most outputs even if they seem very reliable…
Thanks for sharing this, Joel! I enjoy reading such posts that teach us another way of interacting with AI.
100% agree, too much to learn by oneself
NotebookLM is incredible, but it needs time to cook. I load everything first (PDFs, docs, CSVs, images, notes), do a quick citation sanity check, then let it sit 24 hours in the smoker before I ask for Studio outputs. Sometimes 36–48 if the corpus is big 📐
Low and slow 🐢 beats fast and flimsy 🐇
Load → Probe → Cook → Plate
And now I want brisket 🍖😋
haha it does take time! It’s just my research piece, not my whole workflow. That’s claude for me
Thank you Joel for showing the precision side of NotebookLM and how it does not hallucinate by design! The more I feel I have used this tool, the more updates it gets every month! 😁
Thank you, you gave us a very clear strategy on how to implement Notebook LM. It will be very helpful to my personal practice as a teacher and researcher.
Love it, Fernando! Did you get to download the ebook?
Thanks
Hi Joel. Yes, I have. And I am already thinking about ideas to implement it. I just helped a masters student that I advise on creating a notebook to upload all the references and videos he will use in his literature review. I hope Notebook LM is able to read the books in pdf format.
That's great, Fernando. I love to know that these learnings are being passed on. If you ever need anything, please reach out.
Notebook LLM is one of the most powerful AI tools on the market for learning and understanding in my opinion. Thanks for giving us even more tips and tricks for utilizing it to the max!
For sure! I’ll add more as I learn more.
This is such a valuable playbook. NotebookLM is a game changer.
it rlly is!
Source grounded tools change how decisions actually get made.
Exactly. How many people's credibility has been lost because they never went to the source?
THANK YOU, Joel! I did not realize you could directly reference a notebook from Gemini. That is incredibly useful.
It really is! Integration across products
too much fun thank you
Hope it was helpful!
Hmm, wondering if Poppy AI runs NotebookLLM as it's engine?
Poppy AI runs many different ones; it runs some Anthropic, some OpenAI. It’s similar to Notebook LM, but honestly wouldn't surprise me if at some point in the future they are acquired by Google or one of these bigger engines.
This is good, especially the idea that leaders must treat NotebookLM as a research system not a chatbot. That distinction matters when decisions are on the line.
From my experience in delivery and tech leadership, tools that force grounded context and verifiable outputs change how teams converge on strategy - whether it’s onboarding, competitive insights, or product roadmaps.
NotebookLM’s source-grounded approach (where every insight ties back to actual docs) is exactly the antidote to “fluffy confidence” we see with general LLMs.
It’s not just about avoiding hallucinations - it’s about trustworthy operational workflows that scale across teams and reduce risk.
I love the term 'fluffy confidence'!!
Thanks, Joel, for this post. A few takeaways: I haven’t used persona customization in NLM yet, so that was a great idea, and I clearly need to dig more into Studio, I’ve been experimenting with charts but not the more creative parts. Just yesterday I uploaded a 4‑hour YouTube video I didn’t have time to watch, asked for a few key points, and it delivered exactly what I needed. Now I can build on that and add more content to complete my case.
YouTube summaries is one of the main usecases for me as well!