01What Gemini Notebook OS Is and How It Works
Gemini Notebook OS is the engine behind a personal Agent OS that Julian Goldie built to manage all his notebooks, research, chats, and content generation. In this episode, he shows how he turned the raw Gemini Notebook into a one-dashboard system that replaces a chaotic grid of notebooks with a clean library, research tab, chat tab, and studio.
The system works by connecting Gemini Notebook to a custom dashboard using an MCP (Model Context Protocol) bridge. This bridge lets the dashboard talk to the notebook engine and pull everything together. You add sources, the engine reads them, and then you can research, chat, and generate from them—all in one place.
02Main Features of the Agent OS Dashboard
The Agent OS has four main sections. The top left is the dashboard, which shows everything you have built at a glance. The library segment shows every notebook, organized and linked, and you can see which ones have videos, podcasts, or are still in progress.
The research tab is for deep dives, and the chat tab lets you talk to your notebooks directly like a person.
The studio is where the magic happens. From there, you can generate presentations, audio, video, mind maps, infographics, and flashcards—all from your sources. Julian emphasizes that this removes the need to open five tabs or hunt for a notebook from last Tuesday.
It is all in front of you, one system.
03How to Set Up Your Own Agent OS
To set up your own Agent OS, Julian explains that you start with Gemini Notebook, which is free. You also need the MCP bridge, which is also free. If you want the exact system he built, you can get the full Agent OS zip file from the iProfit boardroom, which includes the setup walkthrough.
- Start with a free Gemini Notebook account.
- Get the MCP bridge to connect your notebook to a custom dashboard.
- Install the Agent OS zip file (available in the iProfit boardroom) for a ready-to-go setup.
- Follow the setup walkthrough, even if you are not a coder.
- Connect your sources—articles, notes, or screenshots—into the system.
Once installed, you can start using it immediately. Julian notes that he is not a coder, and neither are his members, so the zip file is designed for easy installation.
04How to Use Gemini Notebook OS for Research and Chat
In the research tab, Julian shows a constant query he runs: asking about the June 2026 releases. The answer is fresh, from the last few days, not weeks old like normal AI chats. There is a toggle for fast and deep.
Fast gives a quick answer, while deep uses a research agent that digs through news and brings back relevant stuff. Julian uses deep when he is building something he will put his name on.
In the chat tab, the custom agent is trained on his sources. So if you ask it about the Agent OS, it answers as if it has read all your notes, because it has. This makes it feel like talking to a personal assistant who knows your entire system inside out.
You can ask anything and get answers based on your own materials.
05Using the Studio to Generate Content in Parallel
The studio is where Julian's jaws-drop moment happens. He types 'create an audio overview of the agent OS' and it starts building a podcast from his own sources. While that is running, he asks for an infographic at the same time.
Parallel agents allow this—you can add a video, a mind map, or a slide deck all at once.
While the agents work, Julian closes the tab and goes to do something else. The agents keep going without him. When he comes back, everything is finished.
This changed his workflow: instead of watching a loading bar, he sets three things running, does real work, and the system delivers. It is like having a small team that works while you are away.
For example, a research report it generated was seven pages, beautifully organized, covering the whole ecosystem, including models used, security setups, and benchmarks. Julian did not write a word.
06Results and Comparisons Shown in the Episode
Julian compares Gemini Notebook on its own to his Agent OS. Raw, the notebook is a mess—notebooks everywhere, no way to know which have videos or what you generated. List view is no better.
He calls it a great engine, but a terrible cockpit. His Agent OS fixes that by providing a dashboard.
He also shows a research report that is seven pages with charts, diagrams, and benchmarks. It covers three big breakthroughs from June, including model sizes, performance, architecture, and safety setups. The infographic uses Google and Nano Banana 2 for visuals, making it look professionally designed.
Julian notes that asking the same question in a normal AI chat gives stale answers from two to three weeks ago, while his system pulls from the last few days.
07Tips, Limits, and Mistakes to Avoid
One key tip is to use the deep research agent when you are building something you will put your name on. Quick answers are fine for casual questions, but deep research ensures accuracy and fresh sources. Julian also warns about building your system welded to one model.
If that model fails, everything breaks. Layered systems keep models as replaceable parts.
A mistake to avoid is watching loading bars. Set multiple agents running in parallel and walk away. Come back to finished work.
Julian also highlights the importance of citations—always click sources to verify, because the worst thing about AI research is not knowing what it made up. With this system, you can check every source.
Limits: building from scratch takes time. But the zip file exists to skip that. Gemini Notebook is free, MCP is free, and you can build a basic Agent OS on free APIs if you want.
08Who It Is For and Why Gemini Notebook OS Matters
This system is for anyone who uses AI for research, content creation, or training. Julian shows use cases like building training material—for example, a flashcard set on AI automation basics for new members. He also uses it for content, turning research reports into video scripts, briefing docs before coaching calls, and mind maps for messy topics.
It matters because it replaces scattered workflows. Before, research lived in 15 places—articles in one tab, notes in another, screenshots in a forgotten folder. Now, you add sources, walk away, and come back to finished work.
It is all cited, so you can trust it. Content is made in minutes instead of afternoons. The whole ecosystem can cost nothing to run, which makes it accessible to anyone.
