Earlier this month, I received an email with half a dozen attached documents, all outlining information on a property I was considering buying.
Pressed for time, I fed the documents straight into Google’s NotebookLM application, which is designed to be a virtual research assistant and is based on Google’s powerful language model, Gemini.
Within a minute, NotebookLM (the LM stands for language model) summarised the highly technical and legal PDFs into a briefing document. I then requested an audio summary. A few minutes later, a recording was available, featuring two AI-generated podcast hosts breaking down the details of the documents into simple layperson’s language.
It meant that when I finally sat down to comb through the documents, I had an idea of the key things to look out for. This is how I now approach any task requiring significant research. I dump everything into NotebookLM and let it do a first pass. It saves me hours of work each week.
Since the debut of ChatGPT in 2022, the world has been flooded with AI-powered tools promising to revolutionise productivity. Yet few have truly delivered on the promise of making artificial intelligence indispensable in daily work and study.
NotebookLM has become the go-to AI tool for me. You can upload Google Docs, PDFs, text files, web links and YouTube videos ‒ up to 300 sources at a time. I’ve been on a mission recently to educate myself on the work of the great philosophers, from Kant to Socrates, Nietzsche to Popper.
Part of that has involved instructing NotebookLM to create tailored podcast episodes, prompting it with instructions for what to include. It has pulled in dozens of sources ranging from The Stanford Encyclopedia of Philosophy to Wikipedia.
An update to NotebookLM recently introduced a game-changing “interactive” mode. You can now interrupt the AI podcast hosts to ask questions, shifting the focus of the conversation to exactly what you need to know, in real time.
For students, researchers, lawyers and any business professionals needing to get up to speed on a topic, NotebookLM becomes a superpower. It’s often a better way of searching for information on a topic than just googling it.
Unlike generic AI chatbots that draw on the entire internet, sometimes leading to hallucinations or irrelevant answers, NotebookLM’s responses are grounded only in the user’s uploaded sources, unless you direct it to websites as well. The caveat is that Notebook LM is only as good as the information you feed into it. If there are errors in the source material, Notebook LM can’t help you, but it is adept at pointing out omissions or flagging where it doesn’t have enough context.
Google claims NotebookLM operates as a closed system, so the documents you upload to it won’t be used to train Google’s models or accessed beyond your own workspace. However, you can share your notebooks with other users, making NotebookLM a useful collaboration tool.
There’s a free version of NotebookLM with limited features. I’m currently paying $37 a month for a Google AI Pro, which includes NotebookLM, two terabytes of data storage and other AI features. It’s a hefty subscription, but as an information worker, it’s proving its worth.
The next phase of these AI assistants will deliver hyper-personalised experiences, predicting user needs based on behaviour and context, and providing real-time, contextual insights. These systems will move beyond reactive information retrieval to proactively highlight knowledge gaps, suggest actions, and even create new knowledge articles or summaries autonomously.
That’s when AI starts to become really powerful. In the meantime, NotebookLM is the best of the current crop of AI tools if you just want to cut to the chase.
