I've always been a forgetful person. As a kid, that meant forgetting my multiplication tables. In high school, that meant spending hours creating flashcards for memorization-heavy courses. Before you blame academics for ruining my memory, I should clarify: I was (and still am) a total nerd. I actually liked learning, and it wasn’t limited to just school. My forgetfulness just refused to cooperate.

A more recent example is Stranger Things. I watched the first season the year it came out, and I’ve had to rewatch it every single time a new season drops, because by then, I have absolutely no memory of what happened. By my fifth rewatch of season one, you’d think I’d have it memorized. I didn't. Instead, it felt like I was watching a new show. Unfortunately, the same thing happens with articles. So I did what most of us do: I started saving everything to a read-it-later app. That worked… until I realized saving something (and even reading it) isn’t the same as actually remembering it.

NotebookLM fixes forgetting by changing how you read

Adding another use case to NotebookLM's growing list

NotebookLM is Google's AI-powered research assistant, and the special part about it is that every instance you create within it (a notebook) is grounded in what you upload. It only knows what you give it, allowing every answer and output it generates to come directly from your sources.

Once your materials are in a NotebookLM notebook, you can chat with them to ask questions and get completely grounded explanations, generate summaries, have it find connections across multiple sources, and even turn your sources into different outputs like Audio Overviews (a podcast-style conversation), Mind Maps, Slide Decks, and Infographics.

One of the sources you can add to a NotebookLM notebook is a webpage URL, which just so happens to be the format most written content is consumed in today. Articles, blog posts, essays, and newsletters are all just URLs. So instead of saving a link to a read-it-later app where it’ll quietly sit in a growing queue, I’ve started dropping it into a NotebookLM notebook, which I like to call the “reading notebook.” This notebook stores the articles currently in my to-be-read queue.

NotebookLM has a feature that lets you select which sources in a notebook are active at any given time. So when I open the reading notebook, and I’m ready to actually sit down with something, I deselect everything else and focus on just that one source. NotebookLM shows the full text in the Sources panel, so I read it right there. But unlike a read-it-later app, I’m not just passively scrolling. As I read, I can ask the AI to clarify a concept, summarize a dense section, or pull out the key arguments. It turns reading into a conversation.

I’ve always been someone who remembers things better when my questions get answered in the moment. When I can pause and ask, “Wait, why does that work?” it sticks.

The second notebook is where remembering happens

The bigger picture makes it stick

The reason I mentioned that NotebookLM lets you select which sources are active is because once I've "read" a specific piece of content the way I just described above, I don't stop there. Instead, once I'm done reading anything in my "reading notebook," I delete it from the notebook and upload it to another NotebookLM notebook I like to call the "archive."

This "archive" notebook holds everything I've read. Immediately after uploading new content to this archive notebook, I begin asking more questions. This time, the questions are geared toward how what I just read connects to content I've already read.

That step is where everything starts to click. Instead of letting an article live in isolation, I'm forcing it to interact with ideas I've already processed. This also means that every time I read something new that's linked to a topic I've read a fair bit about before, and I'm in the archive notebook connecting the dots, I go over the older material in a natural way. So the more I read about a topic, the more I'm inadvertently revisiting and reinforcing what I read before. It's like spaced repetition without the flashcards.

Chrome extensions make maintaining this system a lot easier

The smallest change that made the biggest difference

When I first started using NotebookLM as a read-it-later app, the biggest friction point was getting articles into my reading notebook in the first place. I'd have to stop what I was doing, open NotebookLM, find the notebook, and upload the URL as a source. That's since been solved by Chrome extensions built for NotebookLM.

There are a handful that let you add sources directly to an existing or new notebook with just a few clicks, without even needing to open NotebookLM. That tiny reduction in friction made a bigger difference than I expected. The easier it is to send something to my reading notebook, the more consistent I am about actually using the system.

I finally stopped blaming my memory

For years, I've blamed my memory for forgetting all that I read or watch. And while my memory likely does play a huge part, blaming it wasn’t really doing any good. I needed to build systems around it that enabled me to actually retain, connect, and recall the things I spent time on. The system above allows me to do exactly that.