NotebookLM Review: Where It Helps and Where It Falls Short
Gemini Notebook (formerly NotebookLM) does one thing well: it answers questions from the documents you upload and shows you exactly where each answer comes from. This review covers what that means in practice, the real limitations, and where another tool fills the gaps.
Updated 4 min read
Gemini Notebook (formerly NotebookLM) is a research tool built around one core idea: answer questions from the documents you upload, and cite the exact passage behind each answer. That focus makes it genuinely useful for a specific kind of work — analyzing a fixed set of sources — and less useful for everything else. This review walks through what it does well, where it struggles, and how a tool like fenbs fills the continuity gap that all source-analysis tools share.
What Gemini Notebook does well
Source-grounded answers with citations are the standout feature. When you ask a question, the tool draws on your uploaded documents and shows the source passage behind each claim. This is meaningful for research: you can check whether the citation actually supports the claim, which catches errors that a general AI answer cannot.
The generated outputs — briefing documents, study guides, FAQs, and timelines — are useful starting points. They save time on the first pass through a large document set, identifying what the sources cover and what they emphasize. They are drafts, not finished work.
The Audio Overview feature turns your sources into a two-host AI conversation about the material. It is useful for getting a quick orientation or for listening while commuting, though it is not a substitute for reading the sources yourself.
The source-limit constraint
The free tier limits how many sources you can add per notebook. For small research tasks — a literature review of five papers, a set of lecture notes — this is not a problem. For large research projects with dozens of documents, it forces you to split material across multiple notebooks, which breaks the tool's main advantage: having all your sources in one place.
A paid or Gemini Advanced plan raises the limit significantly. Check the current limits on the Gemini Notebook page (opens in a new tab) before planning a large project around the free tier.
Continuity across sessions
This is the tool's most significant limitation for ongoing work. Gemini Notebook does not remember what you found, what you concluded, or what still needs checking from a previous session. Every chat starts fresh. If you are doing multi-week research, you have to reconstruct your position each time.
This is not a bug — it is by design. The tool is a source-analysis workspace, not a project management system. The gap is real, though: approved findings, open questions, and next steps need a home outside the notebook to survive across sessions.
Storing reference links and work decisions in fenbs closes that gap. The notebook handles source analysis; fenbs holds what came out of it.
Collaboration limits
Sharing a notebook requires the other person to have a Google account and be added to the notebook. That works for a small team but is awkward for external collaboration or for sharing a result with someone who should not see all your working documents. Exporting the generated output — copying the text — is the practical sharing method.
Team research where multiple people are adding sources, tracking decisions, and assigning follow-up work needs more structure than a shared notebook. A dedicated project tool handles roles, decisions, and task assignments in ways a notebook cannot.
How it compares to a general AI assistant
A general AI assistant — Claude, ChatGPT, Gemini — answers from its training data and optional web search. Gemini Notebook answers only from your uploaded documents. The tradeoff is important:
| Gemini Notebook | General AI assistant | |
|---|---|---|
| Source of answers | Only your uploaded documents | Training data, web search, or both |
| Citations | Inline, linked to source passage | Varies; often none or approximate |
| Best for | Analyzing a fixed document set | Open-ended questions, broad research |
| Session memory | None across sessions | Project memory, custom instructions |
| Source control | You choose every source | May include sources you did not choose |
For most research tasks, the answer is: use Gemini Notebook for the analysis phase when you have a defined document set, and use a general assistant for open-ended exploration and writing.
How to keep your research decisions with fenbs
After a research session, the conclusions you reached and the questions still open are the most valuable output. Keep them somewhere reliable so the next session — or a colleague — can continue from a known position.
- Create a free account and add a project for your research topic.
- Save the links to your key source documents in the project's Links and listings section.
- Record approved conclusions as decisions. Record open questions as tasks. Both are readable by any connected assistant.
- Write where the research stands in "Where we left off" before closing the session.
- Connect a supported AI assistant from Integrations to continue in a fresh chat using the project context.
Keep the work that outlasts the notebook
fenbs holds verified source links, approved conclusions, and open questions outside Gemini Notebook. A connected assistant reads them at the start of a new session, so research continuity does not depend on chat history.
Create a free accountFree for one person · No credit card · Works with Claude, ChatGPT, Cursor and more
The honest verdict
Gemini Notebook earns its place for source-grounded research and study. The citation model is genuinely useful and uncommon. Its limitations — no cross-session memory, source-count caps, thin collaboration — are real and specific to how it is designed.
Use it where it is strong: analyzing a document set, generating a first-pass summary, checking what sources actually say. Complement it with tools that handle continuity, collaboration, and follow-up work. See NotebookLM alternatives for other tools in this space.