- Kini AI
- Posts
- The AI That Finally Came Full Circle for Learning
The AI That Finally Came Full Circle for Learning
My Experience with Google's NotebookLM Latest Video Overview Feature + others

Every now and then, a tool comes along that doesn't just feel new, but feels right. I've spent years watching technology try to keep up with the beautifully diverse ways we learn. This week, I had a real 'full circle' moment while giving another test-drive on Google's NotebookLM, and I knew I had to share why. Here's my take on the AI tool that finally feels like it gets us.
As an educator, we talk about VARK learning preferences—visual, audio, reading/writing, and kinaesthetic—and for years, we’ve had to patch together different teaching methods and tools to enable learning for everyone. But with Google’s NotebookLM, I believe we’re seeing a tool that has finally come full circle, elegantly catering to all these learning modes in one powerful package.
So, kini big deal with another AI tool? The magic of NotebookLM is that it’s your own personal research assistant, grounded in your sources. It’s built on a concept called Retrieval-Augmented Generation (RAG), which is a fancy way of saying the AI answers your questions based on the specific documents or sources you give it, not the entire internet. This prevents it from sòsọ kú sò (hallucinating) and keeps the knowledge relevant to you. I call it a "public RAG"—a tool that gives everyone the power to build their own specialized AI, whether you're a student in Lagos, a professional in Nairobi, or a researcher anywhere in between.

One of my notebooks - Anthropic: my dream Company!
The real genius lies in the Studio, which is where the tool truly adapts to how you learn best. Let’s go with the whole learning preference thing - hopefuly, you get why this touches everyone.
For the Visual Learner: The recently launched Video Overview, think of it as an AI-generated PowerPoint presentation, complete with an audio narrative, sometimes like a podcast style, other times its just a single speaker. From my own tests, I've seen a consistent structure: it creates a cover slide, then moves through different layouts—some with charts like bar graphs, others with tables and It can even generate timelines.
This feature seems to be an evolution of concepts from Google Vids, another AI tool for creating presentations, showing how Google is building a connected ecosystem of products. What truly puts the learner in the driver's seat is the ability to customize it. You can prompt the AI to focus the entire video on a specific topic or theme from your sources, which is incredibly powerful for narrowing down a large amount of information.

NoebookLM Video Overview
For the Reading & Writing Learner: This is the foundation of NotebookLM. The chat interface allows for a classic Q&A, while the ability to add your own Notes lets you type out your thoughts and insights, adding them right back into the AI’s context. And of course, you can always upload notes you’ve handwritten and digitised.
NotebookLM Add Note
But it goes further with the Reports feature. Here, the AI can automatically generate study guides, briefing documents, FAQs, and even project timelines from your sources. This is perfect for anyone who needs structured, written material to study from.
NotebookLM Reports
For the Auditory Learner: Forget just reading text. NotebookLM creates an Audio Overview, turning your complex research papers, articles, or textbook chapters into a dynamic, podcast-style conversation between two AI voices. You can even prompt it to focus the discussion on specific themes from your sources. It’s like having personal tutors breaking down ideas for you on your way to Badagry. The best part? The interactive feature allows you to jump into the conversation, ask questions, and get real-time answers, as if you were a guest on the podcast. My guess is that this is likely powered by Google’s efficient Flash models, which are designed for rapid responses. It’s a brilliant application of a true multimodal system—one trained from the ground up to understand text, audio, and images seamlessly.
Audio Overview
For the Kinaesthetic Learner (The "Doer"): That interactive Mind Map is perfect for learners who need to engage and tinker. You can click, expand, and dive into different concepts, actively navigating your knowledge instead of passively consuming it. The entire platform encourages this hands-on exploration. Them finish work here! You can have the Audio Overview playing in the background while you navigate the mind map. This creates a powerful, multi-sensory learning experience where you can hear the concepts explained conversationally while you visually and physically trace the connections between them. It’s a dynamic way to see the big picture and dive into the details simultaneously.
NotebookLM Mind map with Audio Overview
Collaboration and a Connected Ecosystem
Beyond individual learning, NotebookLM is built for teamwork. You can easily share your entire notebook with a study group or colleagues. You have two options: share the full notebook, giving collaborators access to all your sources and the chat history, or share a "chat-only" version, allowing them to ask questions based on your sources without seeing the documents themselves. This makes it an incredibly flexible tool for group projects and shared research.
As you can see, it touches on every possible way we consume and engage with knowledge, putting the learner firmly in the driver's seat. The collaboration feature—allowing you to share your notebook with a study group—is just the icing on the cake.
Google already has a product called Illuminate that lets users choose from different host voices for audio summaries. Bringing that level of choice to NotebookLM—or even allowing us to clone our own voices—would be a massive step toward accessibility for millions of learners.
Rooms for Improvement
However, the tool isn't perfect. I’ve noticed the audio in video overviews can sometimes have sharp, abrupt cuts - honestly, not a big deal for now.
But my biggest thought for improvement comes from an African perspective. For this tool to be truly transformative across the continent, it needs localization to address a major barrier to adoption: accents! (maybe you get it, but not all Africans watch American films). The default American accents in the audio and video overviews can be difficult for many learners to connect with. My key recommendations are to provide users with the ability to select from a library of pre-built local accents (like Nigerian, Kenyan, or Ghanaian) for the narrations.
Google’s other products, like Illuminate, already offer a choice of host voices, proving this is possible (although no African voices). An even more powerful step would be to allow users to clone their own voices, creating a truly personalized and accessible learning experience. This would be a massive step toward making the tool feel natural and relatable for millions of learners across Africa.
But for power users like me or ITKs there might be a problem of source limits (50 for free, 300 for Pro). Although it makes sense to maintain accuracy becasue of context window issues (SKIP THIS IF YOU SABI CONTEXT WINDOW IN AI — it’s a necessary technical safeguard. This is because of what’s known as the AI’s “context window”—think of it as the model’s short-term memory. There’s a finite amount of information it can “look at” all at once to answer a question accurately. The source limit ensures that all your documents can fit within this window, preventing the AI from “forgetting” information and making errors. While Google’s models boast massive context windows (up to 10 million tokens), this cap is a deliberate choice to guarantee the tool remains reliable and doesn't hallucinate—which is the whole point of a RAG-based system.), serious researchers might hit that ceiling quickly.
KINI BIG DEAL (Why Does this matter)
For the first time, a single AI tool has finally come "full circle" by successfully catering to all major learning styles—audio, visual, reading/writing, and kinesthetic—in one integrated package - ALL FEATURES FOR FREE
It's Your Personal, Grounded AI: Unlike general chatbots that can make things up, NotebookLM is grounded only in the sources you provide. This makes it a reliable and accurate research assistant that gives you personalized answers without hallucinating.
It's a True Pedagogical Partner: It doesn't just give you information; it transforms that information into the format that works best for you. It can turn the same set of documents into a podcast (Audio Overview), a presentation (Video Overview), an interactive diagram (Mind Map), or a written report. It respects and adapts to the individual learner, which is a massive step forward for personalized education.
What a time to be a learner. Ultimately, the best tool is the one that adapts to you, not the other way around. NotebookLM is a powerful step in that direction. It’s a reminder that the future of learning isn't just about smarter tech, but about technology that allows you to still be you!
The only question left is: watin you won sabi? Start here: Google’s NotebookLM
I'd love to hear your thoughts. Have you tried NotebookLM? What has your experience been? Hit reply and let me know.
Until next time, stay curious.
Author’s note: This is not a sponsored post, as it expresses my own opinions.
About Me
I'm Awaye Rotimi A., your AI Educator and Consultant. I envision a world where cutting-edge technology not only drives efficiency but also scales productivity for individuals and organisations. My passion lies in democratising AI solutions and firmly believing in empowering and educating the African community. Contact me directly, and let’s discuss what AI can do for you and your organisation
Subscribe to cut through the noise and get the relevant updates and useful tools in AI.
Reply