Image generated by Midjourney
Twenty44 is constantly staying up to date on generative AI tools so we can put the right tools in the hands of our clients. Usually we don’t publish the reviews on our blog, but in the case of Google’s NotebookLM, we’re making an exception.
Here are five first-take highlights on Google’s NotebookLM.
1. Finally a great generative AI tool for people who rely on Google Workspace
If you rely on Google Docs, Sheets, and Slides and store your files in Google Drive, you’ll love how NotebookLM integrates smoothly into the Google ecosystem. No more downloading your Google Sheets, and uploading them into another LLM to work with them. Instead, you can use the documents on your Google Drive and they remain confidential and are not shared or used for model training.
2. It dramatically reduces hallucinations.
One of the biggest challenges with large language models is that they can give you inaccurate responses based on their training data or information they find on the internet. Google NotebookLM doesn’t use internet data. Instead, it uses the documents you share with it and leverages the LLM to extract a response based on your data which dramatically reduces hallucinations and makes it much easier for you to verify its responses when you need to.
3. It's great for handling large amounts of data
NotebookLM can handle up to 50 sources, with each containing up to 500,000 words, allowing users to upload diverse types of content like YouTube videos, Google Docs, and audio files. This makes it extremely powerful for feeding vast amounts of your data into an LLM as reference material.
4. Gives you your own "RAG"
NotebookLM uses something called “Retrieval Augmented Generation” (RAG). In English, that’s like having a super librarian who knows exactly where in your Google documents to look for the most relevant information. When you ask it a question, it doesn’t just look at everything randomly. Instead, it finds the most important parts of your documents and uses them to answer you. This helps make sure that the answers it gives you are accurate and based on your own data.
5. Turns your notes into a podcast.
We can’t talk about this product without mentioning the “Audio Overview” feature. This feature is a “head-snap” because it turns any notes and documents you give it into an audio summary. Imagine two friendly AI voices conversationally discussing the key points of your notes, like a personalized podcast you can easily listen to on your phone or in your car. It’s weird, but for people who process information well by hearing, it’s an amazing way to absorb the information you’ve been working on.
Oh, and it can be hysterically funny. In a Google NotebookLM Audio Summary of a meeting the Twenty44 partners had just prior to our launch, the AI hosts commented on our “Big Launch Energy”. Even better, here’s what it sounds like when someone clever tells the Audio Summary podcast “Hosts” that they are 1) not human and 2) being permanently taken off the air at the end of this podcast.
TL:DR – embedded below is what NotebookLM’s Ai-personalities thought about this blog post.
Ross McLean
Ross is a 25+ year seasoned professional in brand strategy, consumer insight, and creative technology application with a focus on generative AI. Ross has studied AI Leadership & Strategy, Advanced Prompt Engineering and AI-related Change Management at Vanderbilt University and DeepLearning.AI. Prior to his AI- Deep Dive, Ross founded the Over the Shoulder Smartphone Ethnography platform and was EVP and Director of Strategic Planning at Foote Cone & Belding Chicago, leading strategy for brands like Kraft Mac & Cheese and Coors Light. Ross played a pivotal role in FCB’s transformation from traditional to digital.
AI models used in this post: ChatGPT4o with Canvas for spelling and grammar check, Midjourney for featured image, NotebookLM for audio overview.