NotebookLM vs a chat that remembers you

Updated 31 October 2026

Use NotebookLM, now called Gemini Notebook, when the answer should come from your sources. It reads what you upload, answers with citations, and turns the material into study guides, flashcards and audio overviews. Each notebook keeps its own chat history, but it knows nothing across notebooks and holds no picture of you.

Use a chat with memory when you want an AI that knows you. Who you are, what you're working on, what you decided last week and how you like things explained carry from one conversation to the next. Gemini, ChatGPT and Claude each have a memory of their own, and Counterparts, the memory we make, gives Claude a deeper one. For most students and researchers, the answer is both: the notebook for the sources, the chat for you and the work around them.

This is a research summary of Google's, Anthropic's and our own documentation, read on 31 October 2026. An AI drafted this page from the sources below and another AI checked it. No person has run these steps, and nobody has tested the routine further down. Counterparts is ours: Mike LaPeter makes it.

Which tool for which job

What NotebookLM is now

Google renamed NotebookLM to Gemini Notebook on 16 July 2026. Google calls it "the same standalone product", and notebooklm.google now redirects to notebook.google. Notebooks you make there also appear in the Gemini app, with changes syncing both ways, and in AI Mode in Google Search outside the European Economic Area, in English for now.

Google describes it as "an AI-powered research assistant". You add PDFs, websites, YouTube videos, audio, Google Docs or Slides, then chat with them, and each answer comes with "clear in-line citations". From the same sources it makes study guides, briefings, reports, mind maps, flashcards, quizzes, slide decks, infographics, and audio and video overviews.

It's free with a Google account, with higher limits on Google's AI plans. As of late October the free tier allows 100 notebooks and 50 sources in each, and a source can run up to 500,000 words. It runs in the browser, with a mobile app for Android, iPhone and iPad, and your notebooks also open in the Gemini app on Mac.

In its own app, Gemini Notebook answers only from your sources. "Chat responses in Gemini Notebook only use data from your sources," says Google's help, so a creative request like "rewrite the end of my short story" may get "Gemini Notebook can't answer this question". Newer, experimental features can search the web or run code when you ask, and the same notebook opened in the Gemini app or AI Mode can also draw on web search.

Does NotebookLM remember you?

Inside one notebook, it remembers the conversation. Across notebooks, and about you as a person, it doesn't.

Inside a notebook: chat history "is retained and kept private to you", and the chat uses that history to answer. You can clear it from the chat panel's menu. Each notebook also keeps its own chat style, set under Configure Chat, such as a custom "Respond like a PhD student". Chat history stays in the app you chatted in: a chat with your notebook in AI Mode in Search doesn't show up in Gemini Notebook or the Gemini app (Notebooks in AI Mode).
Across notebooks: "Each notebook is independent," says Google's help. "Gemini Notebook can't access information across multiple notebooks at the same time." What you explained in your biology notebook isn't known in your history notebook.
About you: Google says Gemini Notebook uses the files you add, the outputs you generate and your chat history "to build your knowledge base". Its help pages describe no profile of you that follows you between notebooks.

In the Gemini app, a notebook gets a "notebook memory": Gemini can use past chats in that notebook, as long as Keep Activity is on. You can also add an existing Gemini chat to a notebook, where it becomes part of that notebook's knowledge base.

What carries over between notebooks, and what doesn't

Thing	Carries to another notebook?	How to move it
Sources, into a new notebook	Yes, by copying	Copy at the top of a notebook you own, or one shared with copying allowed, copies its sources and Studio content (Create a notebook). Google's FAQ page still says duplicating isn't supported; its how-to page has the Copy steps
Sources, into an existing notebook	No	Add them again. In the Gemini app, a notebook can't use another notebook as a source
Chat history	No	Save the answers you want as notes, then move the notes. Copying a notebook leaves the chat behind
Notes	No	Export a note to Google Docs, then add that Doc as a source in the other notebook. Edits to the Doc don't sync back. Copying a notebook leaves notes behind
Custom instructions	No	Copy them into the other notebook's settings. Within one notebook they sync across Gemini Notebook, the Gemini app and AI Mode
Connected apps	Yes	A newly connected app is on for every notebook; turning it off in one notebook doesn't carry over
Your plan and usage limits	Yes	They belong to your Google account

Two cautions. Deleted notes can't be recovered. And deleting a notebook removes its sources: in Gemini Notebook, any Gemini chats in it go back to your chat list, but in the Gemini app, deleting a notebook also deletes the chats you added to it.

A chat that remembers you

A notebook knows your sources. A chat with memory knows you: your course or field, the argument you're building, the decision you made on Tuesday, and that you like an example before the theory. Each of the big assistants has one now.

Gemini's Memory. With Memory on, "Gemini can learn from your chats to understand more about you and your world." It needs a personal Google account, age 18 or over and Keep Activity on, and it isn't used in Gems or Live chats. In the US, with a paid Google AI plan, you can read and edit a Memory summary. The Google help pages we read don't say whether Memory applies when you chat with a notebook inside Gemini. If you live in Google's apps, this is the memory that fits; Counterparts doesn't work with Gemini.

ChatGPT's memory. ChatGPT has a memory too, which our directory entry describes: saved memories, and a memory drawn from your past chats, managed in its settings. We couldn't download OpenAI's own help page for this one, so check it there before relying on the details.

Claude's memory. Claude's built-in memory is on by default for Free, Pro, Max and Ultra plans. It saves "individual topics as you chat", such as your role, projects, communication preferences and ongoing work, and you can read, edit or delete each topic in Settings. Paid plans can also search past chats.

The closest thing to a notebook inside Claude is a project: a workspace "with their own chat histories and knowledge bases", where you upload documents and set instructions for every chat in it. Free accounts get up to five. Each project keeps its own memory, and uses your memory from chats outside projects too, but not another project's memory (Anthropic's help). So, as with notebooks, what you worked out in your biology project doesn't reach your history project. Health, religion, politics and similar topics aren't saved unless you turn that on.

Counterparts: the memory we make, for Claude

Counterparts gives Claude human-grade memory: it remembers you, your work and itself, so each conversation picks up where the last one left off. Claude writes down what it learned in its own words, as notes rather than transcripts.

What matters sticks, and the rest fades unless it comes up again, so what Claude knows about you stays current.

Each session opens with a short "who I am, who you are, where we left off" that Claude wrote itself, and related memories come up as you talk. Each session becomes a chapter in a journal Claude keeps, and what it believes about you changes slowly, with the older version kept.

It's one memory for your whole working relationship: your thesis, your preferences, what's in progress. In Claude Code and Claude Desktop, what it knows about you reaches every folder you work in, and its notes about a project come up in that project, so the course you started in one place is known in the next. Mike has used it as his daily memory since September 2026.

Where you use Claude decides how you get it:

claude.ai, in the browser or the Claude apps: a hosted beta with nothing to install. It needs an invite today; to ask for one, email mike@resonantventures.com, or join the updates list at the foot of counterparts.ai. Hosted memories are stored on a server Mike runs, with encrypted backups, and what you say in claude.ai goes to Anthropic as usual.
Claude Code on a Mac or Linux, or Claude Desktop on a Mac: a free, open-source install, with no account and no API key. The memories live in one file on your computer, readable only by your account; like anything Claude reads, what a session uses goes to Anthropic. The Claude Code page has the install; for Claude Desktop, the setup is in the Quickstart, including setting the counterparts tools to Always allow. Windows isn't supported yet.
Gemini, ChatGPT or another assistant: Counterparts has no version for them. Use that assistant's own memory, above.

Counterparts doesn't connect to your notebooks. It keeps Claude's own notes about you and your work, not copies of your readings.

How to use both together

Here's a routine we'd suggest for a course or a research project. We haven't run it for this page.

One notebook per course or project. Put that set's readings in it, and ask your questions there, where each answer points to its source.
Keep the answers worth keeping. Use "Save to note" on a good answer, or generate a study guide or report.
Export what you'll need elsewhere. Export the note or report to Google Docs.
Think it through in a chat that remembers you. Bring the exported text into Claude when you need it, and talk about your argument, what still confuses you and what to do next. Claude's memory, or Counterparts, keeps that from one week to the next, so next time you start from where you stopped.
Go back to the notebook to check. Before a claim goes into your essay or paper, ask the notebook to find it in the sources and click the citation.

Privacy, side by side

Gemini Notebook: Google says your content "will not be used to directly train our foundational AI models, unless you choose to provide feedback". Feedback sends the content along to trained reviewers and is kept up to 3 years, disconnected from your account, and it includes any Gemini chats in the notebook, even with Gemini Apps Activity off. On a work or school account, nothing is reviewed by people or used for training.
AI Mode chats with a notebook: these may be used "for ads monetization, personalization, and to improve Google generative AI models". Your sources and instructions are not.
Gemini app chats with a notebook: these follow your Keep Activity setting and the Gemini Apps Privacy Hub. With Keep Activity on, Google can use your chats to improve its services, including its generative AI models, and human reviewers read some of them. Turning it off stops that for future chats, but it also turns off the notebook memory described above. Deleting Gemini activity doesn't delete data in Gemini Notebook.
Claude's memory: you can pause it, reset it, or turn it off for one chat, and incognito chats aren't saved to memory (Anthropic's help).
Counterparts: on a Mac or Linux, one file on your computer, which you can read, export or delete (our privacy page covers the local install). On claude.ai, hosted memories are stored on a server Mike runs, with encrypted backups (hosted section of the privacy page).

Questions people ask

Is NotebookLM the same thing as Gemini Notebook?

Yes. Google renamed NotebookLM to Gemini Notebook on 16 July 2026 and calls it "the same standalone product". The old address, notebooklm.google, now redirects to notebook.google, and your notebooks also appear in the Gemini app (Google's announcement).

Does NotebookLM remember previous conversations?

Inside one notebook, yes: Google says chat history "is retained and kept private to you", and you can delete it from the chat panel's menu (Use chat in Gemini Notebook). It doesn't carry from one notebook to the next, and even copying a notebook leaves the chat behind. For an AI that remembers you across every conversation, use a chat with memory: Gemini's Memory, ChatGPT's or Claude's own, or Counterparts, the memory we make for Claude.

Can NotebookLM use information from another notebook?

No. Google's help says "Each notebook is independent" and it "can't access information across multiple notebooks at the same time". To start a new notebook from an old one, use Copy at the top: it copies the sources and Studio content, but not the chat history or notes (Create a notebook). To move a note, export it to Google Docs and add that Doc as a source in the other notebook.

Does Google use my NotebookLM sources to train its AI?

Not directly, Google says, unless you give feedback: notebook content "will not be used to directly train our foundational AI models, unless you choose to provide feedback". A thumbs up or down sends that content to trained reviewers, kept up to 3 years and disconnected from your account. On a work or school account, nothing is reviewed by people or used for training, even with feedback. Chats with a notebook outside its own app follow that product's rules: in the Gemini app with Keep Activity on, Google can use them to improve its models (Gemini Apps Privacy Hub), and AI Mode chats may be used the same way.

Is NotebookLM better than ChatGPT or Claude for studying?

For studying from your own materials, usually yes: it answers from your sources with citations you can click through, and it makes study guides, flashcards, quizzes and audio overviews from them (Create a notebook). A chat with memory is better at the other half: knowing you, what you've already covered, and how you like things explained, from one week to the next. That's Gemini's or ChatGPT's memory, Claude's own, or Counterparts, the memory we make for Claude. Many students will want both.

Can I use NotebookLM with Claude?

Not directly: the Google help pages we read describe connections to the Gemini app and Search, not to Claude. The practical route is export: save a chat answer as a note, export it to Google Docs (Create and add notes), and paste or attach it in Claude. Claude's own memory, or Counterparts, the memory we make for Claude, then remembers you and the work around it.

Does Gemini remember me across chats?

It can. With Memory on, Gemini learns from your past chats; you need to be 18 or over, on a personal Google account, with Keep Activity on. You turn it on or off under Settings, Personal Intelligence (Google's Memory help). The Google help pages we read don't say whether Memory applies when you chat with a notebook inside Gemini.

How do I get Claude to remember me across conversations?

Claude's own memory is on by default on Free, Pro, Max and Ultra plans and keeps topics you can read and edit in Settings. Counterparts, the memory we make, gives Claude a memory it writes itself, in its own words, across every conversation: it opens each session with where you left off. On claude.ai it's a hosted beta with nothing to install; it needs an invite today, which you can ask for at mike@resonantventures.com. In Claude Code on a Mac or Linux, it's a free install (Counterparts for Claude Code); Claude Desktop on a Mac takes one more step, in the Quickstart on GitHub.

Questions? Ask in r/AIMemorySystems.

Sources

Google's and Anthropic's help pages read on 31 October 2026; OpenAI's help page couldn't be downloaded. These products change often; if something here has gone stale, open an issue.