NotebookLM improves its chat with custom objectives
NotebookLM expands its ability to analyze documents and hold longer conversations with Gemini's 1 million token context window. It also lets you define objectives and roles for each notebook, while adding automatically saved chat history.
NotebookLM can now analyze larger document collections, hold longer conversations, and tailor its responses to the objective you set. Google is rolling out a series of improvements based on the latest Gemini models.
More context for better research
The main update is its analysis capacity. NotebookLM now includes the full 1 million token context window in chat across all plans. In practice, it can work with much more information at once: lengthy reports, articles, transcripts, or several related documents.
Its ability to maintain multi-turn conversations has also increased by more than six times. That means you can ask for a summary, dig into one section, compare two sources, and raise an objection without repeating the entire context in every message.
Google says its tests produced a 50% increase in user satisfaction when responses used a high volume of sources. The company attributes the result to better context comprehension and a broader ability to find connections between documents.
NotebookLM is no longer as limited to answering the literal question. It can review sources from different angles and combine findings to provide a more complete answer, always based on the material you have added to the notebook.
You can define how it should help
The other new feature is already available to all users: each notebook can have a custom objective, voice, or role. To configure it, open the chat settings icon and write how you want it to respond.
For example, you can ask it to:
- Act as an academic advisor who challenges your assumptions and identifies logical flaws.
- Behave like a marketing strategist and deliver a direct action plan.
- Analyze the material from the perspective of an academic, a creative, and a skeptical reviewer.
- Run a text-based simulation with a specific objective and a limited number of decisions.
This changes how you use NotebookLM. Instead of always asking for a generic summary, you can turn it into a study tutor, a report reviewer, or an assistant for preparing a strategy. Quality will depend on the sources and instructions, but you will no longer have to repeat the same approach in every conversation.
History is saved automatically
Conversations will be saved automatically so you can close a notebook and return to it later. You can delete the history at any time, and if you share a notebook, your chats will remain visible only to you. This feature will begin rolling out during the week after the announcement.
For you, the most useful improvement will depend on the work you do: large document collections benefit from the one-million-token context window, while long projects gain continuity from chat history and custom objectives. The important point is that NotebookLM is becoming more like a persistent workspace for research, not just a chat that answers isolated questions.