NotebookLM, now Gemini Notebook, for New Zealand businesses.

What NotebookLM is now called, and what it does

Google renamed NotebookLM to Gemini Notebook on 16 July 2026. It is the same standalone product, existing notebooks carried over, and the old notebooklm.google address now redirects to notebook.google. Notebooks also sync with the Gemini app, and Google says they are coming to AI Mode in Search. Google reports more than 30 million users and over 600,000 organisations.

The idea has not changed. You create a notebook, add sources (PDFs, documents, web pages, transcripts), and ask questions that are answered from those sources rather than from the model’s general knowledge. It can also turn the same sources into Reports, Slides, Audio Overviews, Video Overviews, flashcards and infographics, and can look for new sources through Source Discovery and Deep Research.

What it is good and bad at for NZ teams

It is good at closed sets of documents that people keep asking about: an employee handbook and health and safety policies, a 60-document tender pack, the product manuals for equipment on a farm or factory floor, or a council’s district plan chapters. Because answers are tied to sources, it is easier to check than a general chatbot.

It is weaker when:

• The set is large or moving. Free notebooks cap at 50 sources. Documents that change weekly need someone to replace sources and re-test.
• Access must differ by person. Everyone you share a notebook with sees every source in it.
• The answer must go somewhere. It produces answers and documents, not actions in Xero, ServiceM8 or your CRM.
• Grounded is mistaken for correct. It can still misread a table or merge two clauses, so check the cited passage.

Gemini Notebook plans and limits

Gemini Notebook is upgraded through Google AI plans, a qualifying Workspace plan or Google Cloud. Google’s help centre marks these limits “Subject to Change”.

For businesses, the practical route is Workspace. Google’s NZ pricing page lists “Gemini Notebook with expanded access to features” on Business Standard, at NZ$21 per user a month (marked NZD, GST not stated, billing term to confirm at checkout). A ten-person team on Business Standard pays 10 × NZ$21 = NZ$210 a month for Workspace with Gemini Notebook included. Google says AI Plus, Pro and Ultra are only available in specific regions; check New Zealand availability before planning on them. See our Gemini guide for what else each Workspace plan includes.

Two recent changes matter. Since 2 September 2026, consumer accounts run on a flexible compute budget that refreshes every five hours, based on prompt complexity, chat length, number of sources and features used. Older “50 chats a day” figures no longer apply. And each notebook now has a secure cloud computer that writes and runs code for data analysis grounded in your sources; at launch this went to Google AI Ultra users and Workspace customers with the AI Ultra Access or AI Expanded Access add-ons, with Pro users on the web to follow.

How Google handles what you upload

The account type decides the rules, according to Google’s plan help page:

• Workspace accounts: uploads, queries and responses are not reviewed by humans, even when users give thumbs feedback, and are not used to train AI models.
• Personal accounts: data is not used to train Gemini Notebook unless you give feedback. When you do, Google may review the full context of that interaction, including your queries, uploads and the model’s responses.
• Enterprise via Google Cloud: data stays within your Google Cloud project and cannot be shared externally.
• Public sharing is only enabled for consumer accounts.

The personal-account feedback rule is the one to brief staff on. A thumbs-down on an answer drawn from a client’s file can put that file in front of Google reviewers, which is Google using the information for its own purposes. Under the Privacy Act 2020, that can amount to a disclosure, and as Google has no New Zealand cloud region, an overseas one under IPP 12. On Workspace, Google processing documents for you is a different position, though you still answer for keeping them secure. Personal information in sources, such as CVs or customer complaints, belongs on work accounts only.

Starter project: a staff policy notebook you can trust

This replaces a generic tutorial with a notebook staff will actually use, plus a test that tells you whether to trust it.

1. Gather current policies only. Collect 10 to 30 documents: handbook, leave, expenses, IT use, health and safety. Delete superseded versions first, or the notebook will quote them.
2. Create the notebook on a work account. Name it “Staff policies, October 2026” and add the documents as sources.
3. Set the chat instruction. Paste the instruction below into the notebook’s custom chat settings so every answer follows the same rules.
4. Write 15 test questions. Use questions staff really ask, with answers you know. Include three whose answer is not in any policy.
5. Score the answers. For each, open the cited passage and mark it correct, partly correct or wrong. The three uncovered questions must get “not covered”, not a guess.
6. Share inside the organisation, not publicly. Name an owner who replaces sources when a policy changes and re-runs the 15 questions.

Answer only from the sources in this notebook.
If the sources do not answer the question, reply: "Not covered in our policies. Please ask [name or role]."
For every answer, name the policy and section you used.
Use New Zealand English and keep answers under 120 words.
Do not give legal advice. If a question is about someone's legal rights, point to the relevant policy and suggest talking to [HR contact].

If 13 or more of the 15 are correct and all three uncovered questions are refused, the notebook is ready for staff. If not, the problem is usually the documents: duplicates, contradictions or policies that never answer the question. Our lessons on summarising documents with citations and building a searchable business knowledge base go further.

When to use Gemini Notebook, and when to build something else

Use a notebook when a team needs answers from a bounded document set this week, at no extra cost beyond Workspace. If your documents live in Google Drive, see AI and Google Drive. Compare with Claude Projects or a general chatbot using Claude versus Gemini. Move to a built retrieval system when you need per-user permissions, thousands of changing documents, measured accuracy or answers that feed other software; RAG versus long context explains the trade-off, and our RAG development work starts there.

Limits and failure modes

• Stale sources. A notebook is a snapshot. When a policy changes, replace the source and re-test.
• Audio and video overviews. They are fluent and can still misstate a detail. Review one before sending it to new starters.
• Automatic outputs. Some users see reports or overviews generated automatically when sources are first added. Do not circulate these unread.
• Shared limits. Sharing a notebook does not raise its source cap for collaborators.
• Regional plans. AI Plus, Pro and Ultra are not offered everywhere, so build on Workspace if you need certainty.

Teams that learn to write test questions before trusting a notebook get far more from it; that habit is central to our AI training for business teams.

Written by the Levy Street team in Wellington. Last updated 19 October 2026.