Google's NotebookLM was renamed to Gemini Notebook in mid-2026, but the tool itself is unchanged at its core: a free AI research assistant that answers questions using only the sources you give it. Unlike a general chatbot that draws on the whole internet, it works from the PDFs, documents, and links you upload, and every answer comes with clickable citations pointing back to the exact passage. For students and researchers, this solves the single biggest problem with using AI for research papers: fabricated references. Because the model can only see your uploaded material, it cannot invent a citation that does not exist in your sources.

Step one is getting in. Open your browser and go to notebook.google.com, then sign in with your Google account — the tool is free. Click "New notebook" and give it a clear name, ideally one notebook per project, because notebooks cannot see each other's sources. Step two is adding sources. In the Sources panel, click "Add source" and upload your research papers as PDFs, paste in Google Docs links, add web URLs, or even drop in YouTube links for lectures. The free tier lets you add dozens of sources, and each uploaded file can be very long, so a full set of papers for a literature review fits comfortably. Students assembling an AI study kit will find this pairs naturally with the broader set of free AI tools for students that cover tutoring, writing, and math.

Step three is where the magic happens: ask questions in the chat panel. Instead of reading fifteen papers cover to cover, ask something specific like "What do these papers say about the main causes of battery degradation?" The assistant synthesizes an answer across your sources and puts citation numbers inline. Click any citation and it jumps to the original passage, so you can verify the claim in seconds. This habit — clicking citations instead of trusting summaries — is what separates reliable AI-assisted research from hallucination-prone shortcuts. You can also ask it to compare sources: "Where do Paper A and Paper B disagree?" is a question most researchers dread answering manually.

Step four is choosing a chat style and saving what you learn. The notebook menu lets you switch between a default assistant, a learning-guide mode that quizzes you, or a custom style for a specific tone. Whenever an answer is valuable, click "Save to note" so it becomes a permanent, citable note inside your notebook — and notes themselves can be cited in later answers. Step five is the Studio panel, which turns your sources into study materials: audio overviews that discuss your papers like a podcast, mind maps that show how concepts connect, flashcards, and quizzes. A student preparing for exams can upload a semester's lecture slides and generate a full revision pack in minutes.

For a serious research-paper workflow, organize before you ask. Upload your core papers first, name each source clearly ("Smith_2026_Battery_Study" beats "document_7.pdf"), and group related papers in one notebook. Then run a sequence: first ask for a summary of each paper individually, then ask for the shared themes, then ask for the disagreements and gaps. That last question — "What do these papers fail to address?" — is where a literature review finds its originality. Researchers who also collect information visually can pair this with Google Lens for extracting information from the physical world to digitize printed references.

A few limitations are worth knowing so you do not trip over them. NotebookLM only knows what you upload — it cannot fetch fresh papers or browse the web on its own, so do your literature discovery elsewhere and bring the PDFs here. Its answers are only as good as your sources, and while citations make verification easy, you should still read the key passages yourself before quoting them in your own paper. Finally, treat each notebook as one project: a literature review notebook, an exam-prep notebook, a thesis notebook. Keeping projects separate keeps answers focused, and since it is free, there is no reason to cram everything into one place.