Google's NotebookLM Goes Pro
The popular AI app gains a premium subscription tier, adaptable UI, and interactive Audio Overviews
Google is taking its popular NotebookLM app to the next level. No longer just an experiment, the company is unveiling a series of updates on Friday that position the notebook tool as a fully mature product. The first is a premium tier called NotebookLM Plus, designed for enterprise users. It is followed by a revamped generative AI user interface and an update that lets you interact with Audio Overviews.
Launched a year ago, NotebookLM may seem like an AI-powered version of Evernote or Microsoft OneDrive, but it offers much more. Users can apply artificial intelligence to imported notes and documents to create summaries, outlines, study guides, or other forms of content. It’s handy if you have to manage information spread across many sources within a single notebook.
“It just takes things that often in the past would take 30 minutes or an hour. It gets you a first draft of them in 30 seconds,” Google Labs’ Editorial Director Steven Johnson explains to me in an interview. “We’re seeing anybody who works with knowledge in any form; if you sit down and upload the core documents that are important to what you do into NotebookLM and mess around for 30 minutes, you will see that there’s so much utility there that would be mindblowing if it were two years ago.”
▶️ Click here to read more about Google's latest updates to NotebookLM
Introducing NotebookLM Plus
NotebookLM fans have yearned for higher limits regarding the number of notebooks, Audio Overviews, and chat interactions permitted. To solve this problem, Google is launching the NotebookLM Plus subscription plan. It's available to business and enterprise customers, schools, and universities through Google Workspace or Google Cloud. Anyone who pays for Google One AI Premium will also have access to this premium tier. NotebookLM Plus will be available starting in early 2025.
With NotebookLM Plus, subscribers can now manage up to 500 notebooks and 300 sources per notebook, significantly increasing from the previous limit of 100 notebooks and 50 sources. Additionally, Google has boosted daily usage limits, allowing 500 chat queries and 20 audio generations per day, up from 50 and just three audio generations.
As a paid offering designed for professionals, the company is adding new features. One involves transforming a notebook into a public knowledge base or a Help Center. NotebookLM Plus subscribers can also change the AI's conversational style by choosing a new option called "guide." It's designed to mimic a guidebook or help center and stands in contrast to the default, which is informative and information-rich but "banter-poor." "Guide" is meant to reflect a style befitting a hotel concierge.
Redesigned NotebookLM User Interface
Underscoring NotebookLM’s evolution from a minimal viable product (MVP) to a mature app, Google is giving it a design refresh. The company is moving away from its two-panel design (shown above) and shifting to a three-panel offering. “Think of it as like a flexible or dynamic interface that is adapting the kind of seamless way to whatever you’re trying to do in the product,” Johnson states. And although this UI is new, Google has planted some “seeds” of this look and feel in the past. Nevertheless, I’m told what is rolling out on Friday should be considered NotebookLM’s “foundation.”
The left panel is where you will find your sources. It’s the knowledge you’re importing to accomplish the task you’re working on. This is where the information is used to train the AI. Next to it is the middle panel, where you’ll interact with the AI. Finally, the right panel is a new area that Google dubbed Studio, and this is where you “make things.”
One thing that hasn’t received an upgrade is the model powering NotebookLM. Johnson confirms that Google Gemini 1.5 Pro will still power the app. However, he revealed it would be updated soon to Gemini 2.0, which was released earlier this week.
You Can Interact With Audio Overviews
The final update to NotebookLM introduces Audio Overviews, a feature Johnson says has seen “amazing” adoption since its launch. “We knew it was going to be a hit,” he adds. Users have embraced the tool in diverse ways, from workshopping short stories to importing journals for the AI to narrate their life experiences. Others have used it to have the AI highlight their accomplishments based on their resumes, assist with study materials, and much more. The flurry of activity has convinced Google to double down on the service.
To that end, the company is releasing, in beta, a way to interact with Audio Overviews. First teased at this year’s Google I/O, users can interject through an Audio Overview they created, and the hosts will respond accordingly. “It’s like you’re passing a note to them in the middle of the recording of the podcast,” Johnson states.
This feature will be limited to new Audio Overviews and is limited to English, though there are plans to expand to other languages next year. Google also warns that AI hosts may pause awkwardly before responding or could hallucinate.
And, unlike the rest of NotebookLM today, interactive audio overviews are powered by Gemini 2.0!
▶️ Click here to read more about Google's latest updates to NotebookLM
Today's Visual Snapshot
Hugging Face, the go-to hub for AI models, has released its annual roundup, which showcases the latest trends in open-source artificial intelligence. One standout from the report is this chart: the most downloaded model of the year (see above). In 2024, Alibaba’s Qwen-2.5-1.5B-Instruct was the most sought-after, with over a quarter of downloads. Meta’s Llama-3.1-8B-Instruct-GGUF model was in second with 7.56 percent.
Although the above chart is specific to this year, Hugging Face notes that from 2022 to today, there has been a rise in the preference for smaller models.
Quote This
“When we started Alexa, and we shared that our goal and our mission was to build the world’s best personal assistant, a lot of people scoffed at that. And they scoffed at it because it’s actually really broad surface area and it’s hard to do […] We have a real chance for Alexa to be the leader here. Now, we are in the process right now of rearchitecting the brains of Alexa with multiple foundation models, and it’s going to not only help Alexa answer your questions even better, but it’s going to do what very few generative AI applications do today, which is to understand and anticipate your needs and actually take action for you. So, you can expect to see this in the coming months.”
— Amazon Chief Executive Andy Jassy at the company’s re:Invent conference discussing how Amazon is leveraging AI to make Alexa a more capable and proactive assistant.
This Week’s AI News
🏭 AI Trends and Industry Impact
Inside the launch—and future—of ChatGPT (The Verge)
The AI revolution is running out of data. What can researchers do? (Nature)
Early adopters are ditching Google Search for AI chatbots (Bloomberg Businessweek)
🤖 AI Models and Technologies
Gemini 2.0, Google’s newest flagship AI, can generate text, images, and speech (TechCrunch)
Meta releases Meta Motivo, an AI model to enhance Metaverse experiences (Reuters)
Microsoft’s smaller AI model beats the big guys: Meet Phi-4, the efficiency king (VentureBeat)
Character.ai retrains chatbots to stop chatting up teens (The Verge)
Lambda launches “inference-as-a-service” API claiming lowest costs in AI industry (VentureBeat)
Anthropic’s Claude 3.5 Haiku now generally available (VentureBeat)
Harvard is releasing a massive free AI training dataset funded by OpenAI and Microsoft (Wired)
✏️ Generative AI and Content Creation
Hands-on with Project Astra, Google’s see-all assistant (Axios)
Google’s Deep Research is an AI tool that uses Gemini to do research for you (The Verge)
ChatGPT now understands real-time video, seven months after OpenAI first demoed it (TechCrunch)
Meta debuts tool for watermarking AI-generated videos (TechCrunch)
Google testing Gemini AI agents that help you in video games (The Verge)
Midjourney is launching a multiplayer collaborative worldbuilding tool called “Patchwork” (VentureBeat)
💰 Funding and Investments
Voice AI startup Vapi raises $20 million at $130 million valuation (Reuters)
AI-focused data center startup Crusoe raises $600 million at $2.8 billion valuation (SiliconAngle)
Anybotics raises $60 million to bring more autonomous industrial robots to the U.S. (TechCrunch)
Software testing platform LambdaTest secures $38 million for AI push (TechCrunch)
☁️ Enterprise AI Solutions
Google unveils AI coding assistant “Jules,” promising autonomous bug fixes and faster development cycles (VentureBeat)
ServiceNow open-sources Fast-LLM to help enterprises train AI models 20% faster (VentureBeat)
How Databricks is using synthetic data to simplify evaluation of AI agents (VentureBeat)
Stainless helps build software development kits (SDKs) for OpenAI, Anthropic, and Meta (TechCrunch)
⚙️ Hardware, Robotics, and Autonomous Systems
Cruise’s robotaxi service to likely shutter after General Motors announces it is pulling its funding (The Verge)
Google unveils Trillium AI chip that delivers 4x speed and powers Gemini 2.0 (VentureBeat)
Proto Hologram’s holographic AI avatars: “A chatbot in a fancy dress, but with a persona presence” (My Two Cents)
Dr. Rob’s new AI model promises to cut aircraft design time from months to days (The Next Web)
🔬 Science and Breakthroughs
💼 Business, Marketing, Media, and Consumer Applications
Google CEO Sundar Pichai jabs at Microsoft’s AI efforts: “I would love to do a side-by-side comparison of our models. They’re using someone else’s models.” (Windows Central)
Founder who built Snap’s AI launches eSelf, a platform for building and operating real-time, video-based chatbots (TechCrunch)
The secret to AI profitability is hiring a lot more doctorates (Bloomberg)
⚖️ Legal, Regulatory, and Ethical Issues
Helen King’s job is all about keeping AI safe as Google scales (Fast Company)
Mexico is using an AI-powered app to prevent suicides (Rest of World)
Character AI hit with another lawsuit over allegations its chatbot suggested a teen kill his parents (Business Insider)
AI’s hype and antitrust problem is coming under scrutiny (MIT Technology Review)
💥 Disruption, Misinformation, and Risks
Microsoft Recall screenshots credit cards and Social Security numbers, even with the “sensitive information” filter enabled (Tom’s Hardware)