Building a Second Brain in NotebookLM Around Your Sources

Building a second brain in NotebookLM starts with one focused notebook, a small collection of relevant sources, and questions whose answers you can check against those sources. Save the conclusions that are useful, together with the evidence and limits behind them. Google’s current help pages use the name Gemini Notebook; the steps here follow that documentation.

A research notebook helps you understand material. When that understanding creates tasks or approved project choices, fenbs can keep the work record available to different connected assistants. Keep the notebook as the research collection and carry forward the reviewed context that the next session needs.

How do you start with a useful notebook?

Choose an outcome narrow enough to guide what you include. Preparing a proposal for a community garden is a workable notebook topic. Everything about gardening is likely to become a collection you cannot sensibly review. Write the question you want to answer before adding documents.

1. Create a notebook for one project or research question. Give it a name that explains the intended output.
2. Add a few authoritative sources you are allowed to use, such as the project requirements, a relevant guide, and your own approved notes.
3. Ask a narrow question and inspect the source passages behind the answer before saving a conclusion.
4. Save the useful answer or write a reviewed note, keeping your interpretation distinct from the original evidence.
5. Record the next action and the remaining question so the notebook helps you resume the work.

The AI second-brain introduction explains the wider idea of reusable knowledge. You do not need to reproduce a complete personal filing system inside this first notebook. A small collection you understand is a stronger starting point than a large collection whose relevance is unclear.

Which sources should you add first?

Start with the evidence that constrains your answer. In the garden example, that could mean the actual site requirements, the application instructions, and a planting guide relevant to your area. Add background reading only when it helps answer a question that the core material leaves open.

Google’s source documentation↗ (opens in a new tab) lists supported material including Google Docs, PDFs, web pages, pasted text, and other file types. It also explains selecting sources for a particular question. Check that documentation for current formats and limits rather than designing your system around a fixed allowance.

• Use descriptive source titles so you can identify the document in an answer.
• Keep the original link, author or organization, and publication or revision information where available.
• Distinguish official requirements from examples, commentary, and your own working notes.
• Check the original when a rule or specification may have changed, including how that source type updates in the notebook.

Do not treat adding a source as approving its contents. A draft proposal and a final requirement can disagree for a good reason. Explain their roles in your question so you are asking for a comparison, rather than an answer that silently blends them.

What questions produce answers you can check?

Ask about a decision you actually need to inform. Which of our proposed beds satisfy the application’s size requirements? gives the notebook something concrete to locate. Make the selected source set fit the question, especially when you have drafts or superseded guidance in the collection.

Use the application instructions and our draft proposal.
List each stated requirement relevant to the planting plan.
For each requirement, cite the supporting source passage.
Say whether the draft addresses it, misses it, or is ambiguous.
Do not invent a requirement when the sources are silent.
Separate your suggested improvements from stated requirements.

Google’s chat guidance↗ (opens in a new tab) explains that citations let you inspect supporting source material, including opening a quote in context. Use that route for important statements. A citation is a place to check the answer, rather than a reason to skip the check.

Compare the claim with the surrounding passage. Does the rule apply to this project? Is there an exception? Does a suggestion in a source appear in the answer as a requirement? Resolve those questions before turning the answer into project instructions.

How should you save conclusions as notes?

Save a note when you can explain why you will need it again. A useful note captures the conclusion, the source it depends on, and what remains uncertain. It should let your future self understand the reasoning without rereading a long chat.

Google’s notes documentation↗ (opens in a new tab) describes writing notes and saving chat responses. It also explains that selected note sources can be used in later answers. That makes it important to label your own conclusions clearly: a later citation to your summary is not an independent check of the original evidence.

Question: Does our draft address the site requirements?
Conclusion: [what the sources support]
Evidence: [source title and passage location]
Limits: [what is uncertain or outside the source’s scope]
Checked by: [person and date]
Suggested next action: [work still to do]
Approval state: [open or the human choice actually made]

When a conclusion affects active work, put a concise checked version and its reference in fenbs so the next connected assistant can find it through the project context workflow. Keep research suggestions distinct from choices a person has approved.

How to do this with fenbs

1. Create a free account and add the project your notebook supports.
2. Add the source references and a short context note with the conclusion you checked. Include the evidence limit the next assistant must respect.
3. Record choices a person made in Decisions and rules. Add tasks for remaining work, with a clear plan and any checks still needed.
4. Connect a supported assistant through Integrations. Ask it to read the project context before working and save progress and Where we left off before stopping.

fenbs does not import NotebookLM chats or provide a direct notebook connection. It is not the right fit for questioning a large source collection inside a research notebook. Its role is the ongoing work record; the AI brain guide explains what belongs there.

How do you keep the notebook useful over time?

Review the source set when the project changes. Remove an irrelevant draft from the question you are asking, identify material that needs a fresh check, and rewrite a conclusion when the evidence changes. Retaining an old note without its original context can make a once-useful answer misleading.

Test the system with a question you know well. Can you locate the right passage, see the difference between a requirement and a suggestion, and identify the next action? Then test a question the sources cannot answer. A useful setup makes that gap visible rather than covering it with a confident summary.

The portable AI memory guide covers keeping work available when you change assistants. For a system outside a research notebook, see build your own AI second brain. The second-brain topic hub gathers other approaches. Keep this notebook’s purpose concrete: helping you reach a supported conclusion for the project in front of you.