We've reached a stage where a lot of the features AI tools are launching feel incredibly similar. Now, given that all of these tools are competing with each other, it only makes sense — chasing the same set of features is inevitable when everyone's trying to simplify the same use cases. That said, NotebookLM is one of the few tools that genuinely excites me every time it launches something new. I've always loved its vision of working with your own sources, and every new feature it ships is built directly on that foundation.

This means that everything you create from the tool stays grounded in your actual material, and that alone makes its features worth using in a way that most AI tools struggle to match. The features are also always unique, and they work well right out of the gate. In the beginning of March, NotebookLM launched a feature called Cinematic Video Overviews, and I frankly don't think I've ever refreshed an AI tool so many times hoping to see a single feature drop. I finally got access to it a couple of days ago, and while I'm a fan, it comes with a huge caveat.

What is NotebookLM's Cinematic Video Overviews feature?

Audio Overviews walked so Cinematic Video Overviews could run

The feature that made NotebookLM go viral on social media and essentially began its boom was Audio Overviews. This feature lets you convert sources uploaded in a notebook to full-fledged podcasts, and the impressive part about them is that they sound incredibly natural. There's banter between the hosts, they're fun to listen to, and they make for a great passive learning experience. At Google I/O 2025, the company decided to double down on the success of this feature by adding a video element to it, launching Video Overviews. The concept behind it is just what you'd expect — generate a narrated video rather than just audio playback.

I was beyond excited for this feature, but it was honestly disappointing. The videos you could generate were essentially narrated slideshows. They got the job done, but the visual experience was pretty flat. One slide would move to the next, an AI host would narrate the content, and that was really it. It felt like you were watching someone narrate a PowerPoint presentation, and nothing more! Cinematic Video Overviews is NotebookLM's answer to that gap.

Rather than generating narrated slides, Cinematic Video Overviews lets you create fully immersive videos. It achieves this by using a combination of Google's AI models: Gemini 3, Nano Banana Pro, and Veo 3. This means that unlike standard Video Overviews that have mostly static elements, a Cinematic one you generate will have fluid animations and detailed visuals.

The feature is a dream for visual learners like me

My YouTube watch history is about to get a lot shorter

My biggest use case for NotebookLM has always been studying, and when it comes to understanding confusing concepts, I've always preferred videos over text explanations. I spend hours on YouTube watching videos to clear up confusing concepts, understand different perspectives on a topic, and just generally absorb information in a way that actually sticks. This is exactly why I was so excited for the Video Overviews feature when it was announced at I/O. I've spent an embarrassing amount of time actually browsing through YouTube trying to find the right video to watch that would actually explain the exact concept I'm trying to understand. But of course, finding one video that includes the exact information you need, at the right level of depth, without a ten-minute intro or a bunch of other content jammed in, is rarely easy.

Since NotebookLM only generates content from your own sources, I knew Video Overviews would solve that problem. I already use the tool to study and understand concepts, but it's always been me asking questions, getting answers, and then having back-and-forth conversations. But after generating multiple standard Video Overviews, they just weren't engaging enough to keep me interested, and I turned back to YouTube. Cinematic Video Overviews changed that completely. The fact that it doesn't feel like you're watching someone narrate a script over a set of slides makes all the difference. The animations, the pacing, the visuals — it actually holds your attention in a way that makes you want to keep watching. And of course, it isn't just limited to studying!

For example, I was recently working on a medical insurance-related project for a consulting company, and since I'm in no way part of the insurance world, I was incredibly confused by pretty much everything — the terminology, the workflows, how verification processes actually work. I had a notebook full of sources about insurance eligibility verification, and honestly, reading through it all felt overwhelming. So I generated a Cinematic Video Overview from it, and it genuinely made the entire thing click. Watching the concepts play out visually, with the narration walking me through each step, was so much more engaging than staring at walls of procedural text.

Rather than the eligibility steps written on a slide and the host explaining them, the Cinematic overview actually animated the entire verification flow. The best way I can describe the experience is that it feels like someone is sketching things out on a whiteboard right in front of you.

The limits are absolutely atrocious

Two videos a day. Two.

AI tools having brutal limits has been a huge topic of conversation lately. Google genuinely disappointed me with how bad the limits of this feature are. It's not available on the free tier yet, and only rolled out to Google AI Pro a few days ago. It was initially only available on the Google AI Ultra plan, which costs $250/month! That's precisely why it's taken me so long to test it out. On top of that, AI Pro subscribers can only generate two Cinematic Video Overviews per day. Two!

The first time I finally began testing the feature, the first video I was generating failed midway, and I was left with only one more video to generate for the day. That means one failed generation eats into your already tiny limit, and there's no way to get it back. For a feature this impressive, having such a restrictive cap makes it genuinely hard to experiment with — and experimenting is exactly what you need to do to get the best results out of it, as with any AI feature.

You can't iterate on your steering prompts, try different source combinations, or compare visual styles when you're constantly worried about burning through your daily allowance. And while the feature is impressive, it isn't impressive enough for me to justify paying $250 to generate 20 videos per day.

Limits aside, this might just be my favorite NotebookLM feature

As a visual learner, this is genuinely one of my favorite features NotebookLM has launched in the last few months. I'm excited to see how they improve it, and have my fingers crossed that Google loosens up the limits as the feature matures.