01What Is NotebookLM AI Agent?

NotebookLM AI agent is a free research engine that helps you turn your ideas into structured reports, podcasts, videos, and more. It works by letting you plug in sources, then it reads them, connects the information, and answers with citations.

In this episode, Julian Goldie explains three big updates to NotebookLM. First, a better chat experience with curated skills. Second, you can visualize data, PDFs, images, and Excel sheets.

Third, there is an agentic research companion that guides you through your ideas.

You can use it for deep research, generating content, or even asking questions and getting cited answers. It is like having a custom AI agent trained on your own sources.

023 Big Updates in NotebookLM AI Agent

The first update is a better chat experience. You can plug in sources, add your own, and access over 100 curated software skills. This gives you deeper research and more complex analysis.

The second update is data visualization. You can now visualize data, PDFs, images, and Excel sheets directly in NotebookLM. This makes it easier to understand your information.

The third update is an agentic research companion. You start by entering loose ideas and questions into the chat. Then NotebookLM guides you through the process, helping you get to a refined result.

03How to Set Up NotebookLM AI Agent

  1. Go to NotebookLM and create a new notebook. You can type in a new notebook name in the library.
  2. Add your sources. You can plug in links, PDFs, or other documents.
  3. Use the research section to ask questions. You can switch between fast and deep research modes.
  4. Use the chat to ask general questions and get cited answers.
  5. Use the studio to generate outputs like podcasts, videos, infographics, or slide decks.
  6. Optionally, build an agent operating system using an MCP to organize everything, or join the AI Profit Boardroom to get a ready-made system.

04How to Use NotebookLM AI Agent for Research

To use it for research, feed it with sources like links or documents. Then ask a question. NotebookLM will read the sources and give you cited answers.

For example, Julian asked: "What is the one thing a builder should know about the June 2026 release?" It came up with answers and sources, including the new Cohere North Mini code update, GLM 5.2, and Hermes agent. All this news was from the last few days, unlike other tools that show outdated info.

You can also go to the research section and choose deep research. This uses the research agent to look deeply and find relevant information. After research, you can generate a report, podcast, or infographic from the studio.

05Results and Tests in the Episode

Julian showed a research report that was seven pages long. It was beautifully organized with diagrams and charts. It covered the ecosystem, setup, loops, and models used.

He also showed an infographic generated by NotebookLM. It looked amazing because it uses Google's nano banana 2 model. The slide deck also looked great.

He compared asking the same question in Claude or similar tools. They gave outdated information from weeks ago. NotebookLM gave news from the last few days.

This makes it more powerful for staying current.

06Tips, Limits, and Mistakes to Avoid

  • Avoid letting NotebookLM get messy. Julian says notebooks can be all over the place, so use an agent operating system to organize them.
  • Don't rely on tools that give outdated answers. Use NotebookLM because it gives recent, cited sources.
  • Don't think it's too technical. Julian says he is not a coder, and neither are his members, yet they build these systems.
  • Use parallel agents. You can generate multiple things at once, like podcasts and infographics, saving time.
  • Be aware of limits: NotebookLM's native interface lacks a clear library, so use an agent OS for better organization.
  • Avoid building a system where you can't swap models. Julian mentions Fable 5 getting taken down by Claude, so having a modular system gives you control.

07Who Is NotebookLM AI Agent For?

NotebookLM AI agent is for anyone who does research or creates content. It is especially useful for people who need cited answers and up-to-date news. Julian uses it to train his team, create content, and generate social media posts.

It is also for people who want to build a free, powerful research system. NotebookLM itself is free, the MCP is free, and you can use free APIs. You can build the whole ecosystem for free, or join the AI Profit Boardroom to get the agent operating system.

If you are not technical, you can still benefit. Julian says his members, who are not coders, get amazing results. With a community, you can learn and grow together.

08How the Instant Research Engine Works

The instant research engine uses NotebookLM as the power. You feed it a pile of links or sources. It reads them, connects the information, and then performs by generating assets.

You can generate a podcast, a briefing document, a mind map, flashcards, or a study guide. All inside one organized system. Julian showed how it turns research into stuff you actually use.

For example, he fed it information about his agent operating system. Then he got a podcast, an infographic, and a slide deck. The quality looks nice because it uses Google and nano banana 2 for outputs.

This saves time. Instead of having articles and notes everywhere, you add sources, walk away, and come back to finished content.