If you use Google Gemini or NotebookLM regularly, the latest updates are a big deal. These are not tiny interface tweaks or background improvements nobody notices. Google has added real customization to NotebookLM, made Gemini more context-aware, improved image generation, expanded file and media support, and pushed its creative tools much further with Google Flow and geo-grounded visuals.
The biggest theme across all of it is simple: more control. You are no longer stuck with one default output and forced to accept whatever the system gives you. You can shape the result, refine it, and build with much more precision. That changes how useful these tools are for research, studying, content creation, and visual work.
If you want to get the most out of Google Gemini and NotebookLM, these are the updates worth paying attention to.
Tags: Google Gemini, NotebookLM, Gemini updates, NotebookLM updates, Google Flow, AI image generation, Nano Banana 2, AI productivity
NotebookLM now gives you far more control over what it creates
One of the most important updates is that NotebookLM is now much more customizable across the board. That may sound small at first, but it fixes one of the biggest frustrations people had with the product.
Previously, a lot of generated outputs felt fixed. You could ask NotebookLM to create something, but you had limited control over how that thing was shaped. Now that has changed.
You can customize outputs like:
- Flashcards
- Reports
- Slide decks
- Infographics
- Other generated study and content assets
Flashcards are no longer one-size-fits-all
Flashcards are a great example of how much more flexible NotebookLM has become. You can now decide things like:
- How many cards to generate
- How difficult they should be
- What range of material they should cover
- Whether they should include even the smaller details
That matters because not everyone studies the same way. Sometimes you want a lighter review set. Other times you want hard questions that dig into specifics. Now you can steer the tool instead of treating it like a vending machine.
In one example, the system generated dozens of flashcards from source material, showing that NotebookLM can build a substantial study set while still letting you control the depth and style of the result.
Reports can be shaped for different end goals
Reports are also much more adaptable now. Instead of getting a generic output, you can choose a format that actually matches what you need.
That includes options such as:
- Briefing document
- Study guide
- Blog post
NotebookLM can even suggest output directions based on the sources inside your notebook. So if the material leans toward strategy, operations, or frameworks, it can surface suggestions that fit that topic.
This makes NotebookLM feel less like a summarizer and more like a tool that helps package information for a real use case.
You can now inspect and iterate on NotebookLM outputs
Another major improvement is transparency.
When NotebookLM creates something, you can now dig into the details behind the output. That includes:
- The prompt used to generate it
- The source material involved
- The ability to iterate and revise the prompt
This is huge.
If the first result is not quite right, you do not have to start over from scratch or just hope the next try is better. You can inspect what happened, tweak the instructions, and generate a more targeted version.
That turns NotebookLM into a much more serious workflow tool. It also gives you a better understanding of why an output turned out the way it did.
Sharing and repurposing are easier too
Generated assets can also be shared more easily. For example, flashcards can be shared with other people, and outputs can be turned into artifacts for broader use.
That is useful for:
- Study groups
- Team documentation
- Research collaboration
- Content repurposing
Gemini gets better when location is turned on
Google Gemini now becomes more relevant when location access is enabled on the device you are using. This is one of those updates that some people will love and others will skip, but if your use case involves local recommendations or location-specific answers, it can make a real difference.
For example, this can improve responses related to:
- Restaurants
- Hotels
- Local businesses
- Area-specific planning questions
Of course, privacy preferences matter. Not everyone will want to enable location. But if local accuracy is important to you, it is worth considering. Just remember that this setting needs to be enabled separately on each device where you use Gemini.
Gemini image generation is getting a serious boost with Nano Banana 2
Google has also updated image generation with a model referred to as Nano Banana 2, and the improvements are noticeable in two areas: speed and editing quality.
This model makes it easier to:
- Edit existing images
- Create new images
- Apply visual transformations
- Experiment with different styles and materials
One example is turning a normal photo into a marble sculpture. It is the kind of transformation that shows how quickly Gemini can now restyle images while preserving the main subject.
That said, the model is only as good as the instructions you give it.
Why better prompts still matter
If you type something vague like “Ferrari F40,” you might get a result, but it may not be the image you actually want. The problem is not always the model. Often the issue is that the prompt is too weak.
To get a much better result, the prompt needs more structure, including details like:
- Subject and scene
- Style or artistic direction
- Composition and framing
- Lighting and mood
- Technical quality expectations
- Negative constraints for things to avoid
That is where prompt optimization tools can help. A Chrome extension called MyPromptBuddy was highlighted as a way to improve image prompts directly inside Gemini. The idea is straightforward: instead of tossing in a lazy prompt and then wasting time on retries, build a stronger prompt up front.
The difference can be dramatic. A weak prompt often produces a generic or disappointing image. A structured prompt can turn that into something polished and far more professional.
If you are doing any serious AI image work, this is the lesson: better prompting still wins.
Gemini now has a cleaner library and clearer usage limits
Another practical improvement is better organization.
Inside Gemini, the library area now makes it easier to find generated documents and assets. This is one of those quality-of-life updates that matters more the longer you use the product. Once you start generating a lot of content, having a clean way to revisit it becomes essential.
Google has also made usage limits more visible in settings. You can now see:
- Your daily usage status
- Your weekly usage status
- Upgrade options if you need more capacity
That transparency helps you plan around limits instead of running into them blindly.
Gemini’s upload options and context tools are expanding
When starting a new chat in Gemini, the plus menu now supports more ways to add context. This makes Gemini more useful because strong outputs depend heavily on the quality and range of the information you provide.
New and improved options include:
- Turning personal intelligence on or off
- Adding files from Google Drive
- Using more upload types
- Pulling in content from Google Photos
- Using materials from notebooks
This makes the overall experience feel more connected. Instead of treating Gemini like an isolated chat box, Google is turning it into a workspace that can pull in relevant context from across your ecosystem.
Gemini 3.5 Flash adds computer use capabilities
For developers, one of the more important announcements is the addition of computer use capabilities with Gemini 3.5 Flash.
This pushes Gemini into a more agentic direction. In simple terms, it means the model can interact with digital environments and perform actions that resemble actual software use.
That includes the ability to work inside app environments and make use of mobile-style capabilities. This is less about casual chat and more about building tools, workflows, and autonomous interactions.
For most people, this update matters as a sign of where Gemini is heading. For developers, it matters because it opens the door to creating systems that do more than just answer questions.
The setup can be done through Google AI Studio, where the implementation is designed to be relatively straightforward for those building with the platform.
If you work in development, automation, or AI product design, this is an update worth exploring further through Google’s official documentation and tools at Google AI Studio.
Google Flow is becoming one of the wildest creative AI tools in the stack
The most mind-bending update of the bunch may be what is happening inside Google Flow.
Google Flow now supports Maps Imagery Grounding, which means scenes can be generated using Google Maps Street View as a real-world reference. That gives your AI-generated images and videos a much stronger sense of place.
Instead of making a totally invented city block, you can ground a scene in an actual location and then creatively transform it.
That opens up a lot of possibilities:
- Placing characters in real neighborhoods
- Stylizing known landmarks
- Reimagining familiar places
- Creating more believable cinematic scenes
- Building documentary-style or narrative visual concepts
A simple prompt can produce something cinematic
One example used a real location, Main Street in Bozeman, Montana, and reimagined it with giant ants roaming around. The interesting part is not the ants. It is the workflow.
Google Flow can:
- Pull in grounded location imagery
- Use an image model to build realistic scenes
- Create characters and visual elements
- Combine everything into a coherent composition
- Turn those scenes into video if needed
That is a big step forward. It gives creators a way to blend fantasy with geographic realism, which can make AI-generated visuals feel more convincing and more useful.
Google Flow’s interface is becoming much more capable
The toolset inside Google Flow has also expanded. You can now work with:
- Image creation tools
- Video creation tools
- Prompting tools
- Custom creative tools
- Scene management
- Character creation and uploads
- Model selection defaults
- Generation preferences
There is even a section for agent instructions, which acts like a persistent skill layer. If you want the system to repeatedly follow certain creative rules or behaviors, you can define those once and reuse them.
This is the kind of feature that starts to matter when you move beyond one-off experiments and begin building repeatable creative workflows.
More NotebookLM and Gemini features are on the way
Some of the most interesting changes have not fully landed yet, but they show where Google is heading.
NotebookLM is getting deeper personalization
Upcoming NotebookLM improvements include AI-assisted editing descriptions and stronger personal preferences. In practice, that means the system will better understand how you work over time by using signals such as:
- Past conversations
- Previous artifacts
- Custom instructions
- Interaction history
This points toward a NotebookLM that feels less generic and more adapted to individual workflows.
Gemini on macOS is moving toward voice-driven productivity
Google is also preparing a Gemini desktop app for macOS with a voice dictation feature called Speak to Window. The idea is that you can use voice commands to direct Gemini while continuing to work in another app.
That could enable tasks like:
- Drafting emails
- Writing documents
- Creating images
- Managing work across windows without constantly switching context
If this works smoothly, it could become a very practical productivity feature.
What these updates actually mean
All of these releases point to the same broader shift. Google is not just making AI models smarter. It is making the surrounding products more usable, more customizable, and more integrated into real workflows.
That matters because raw model quality is only part of the story. A tool becomes valuable when you can:
- Control the output
- Understand how it was created
- Reuse it across projects
- Feed it richer context
- Generate text, images, and video from the same ecosystem
NotebookLM is getting better at turning source material into useful assets. Gemini is getting better at handling context, media, and structured interactions. Google Flow is becoming a serious creative playground with real-world grounding.
Taken together, these are not random updates. They are signs of a much larger platform maturing quickly.
Suggested media to include in this article
- Screenshot of NotebookLM flashcard customization interface with alt text: “NotebookLM flashcard customization options for card count, difficulty, and coverage”
- Screenshot of Gemini image generation results with alt text: “Google Gemini AI image generation example using Nano Banana 2”
- Screenshot of Google Flow maps grounding interface with alt text: “Google Flow Maps Imagery Grounding for AI scene creation”
- Infographic comparing old versus new NotebookLM customization options with alt text: “Comparison of NotebookLM customization features before and after the update”
FAQ
What is the biggest NotebookLM update?
The biggest change is customization. NotebookLM now lets you shape outputs like flashcards, reports, slide decks, and infographics with much more control over format, depth, and difficulty.
Can you edit NotebookLM outputs after they are generated?
Yes. You can inspect the prompt and sources behind an output, then iterate by changing the prompt to improve the result.
What is Nano Banana 2 in Google Gemini?
Nano Banana 2 is the image generation model mentioned in the update. It improves both image editing and generation speed, making visual creation inside Gemini more capable and efficient.
Does turning on location improve Gemini responses?
Yes, especially for local questions such as restaurant suggestions, hotel recommendations, or area-specific information. The setting must be enabled on each device separately.
What is Google Flow’s Maps Imagery Grounding?
It allows Google Flow to generate visuals based on real-world location imagery from Google Maps Street View. This helps create scenes that feel more realistic and geographically accurate.
Who benefits most from Gemini 3.5 Flash computer use features?
Developers and builders will benefit the most, since the feature is designed for app interaction, automation, and agentic workflows inside development environments like Google AI Studio.
Final thoughts
If you have not checked in on Google Gemini or NotebookLM recently, now is the time. The latest changes make both tools more practical, more powerful, and a lot less rigid.
The real win here is not just that the models can do more. It is that you can now guide them better. That is where productivity jumps. That is where better outputs happen. And that is where these tools start moving from novelty to something you can actually build around.
If you are experimenting with these updates, test the customization features first. That is where the biggest immediate gains are likely to show up.
