The NotebookLM Update Most Teachers Cannot Access Yet

Google upgraded NotebookLM's underlying model to Gemini 3.5 on June 8, 2026, alongside a set of new features significant enough that the update has been widely covered as one of the biggest changes to the platform since its early days. For teachers who have built the unit notebook workflow covered elsewhere on this site, the real question is not whether this update is impressive in general. It is whether any of it actually changes how a teacher would use NotebookLM for a classroom unit, and whether the new features are even available on the access level most teachers actually have.
That second part matters more than most coverage of this update mentions, so I want to address it directly before getting into the features themselves.
What Actually Changed
The June 8 update moved NotebookLM onto Gemini 3.5 as its default underlying model, which improves the quality of the reasoning and analysis NotebookLM can perform across your uploaded sources. Alongside the model upgrade, three specific new capabilities were introduced.
Notebooks can now build their own source list through conversation. Previously, using NotebookLM required you to gather and upload your sources before you could start querying them. The update introduces a guided discovery feature where you can describe what you are researching in chat, and NotebookLM will suggest sources using its own research capabilities and Google Search, helping you build out a source collection rather than requiring you to arrive with one already assembled.
Each notebook now includes what Google describes as a secure cloud computer, allowing NotebookLM to write and run code for deeper analysis of the sources in a notebook. This is aimed primarily at data analysis tasks rather than typical classroom research use, though it has some relevance for any teacher using NotebookLM with data sets, such as a science class analyzing experimental results.
Output formats have expanded beyond the audio overviews and study guides NotebookLM already offered. The update adds the ability to generate charts, structured data tables, and files in formats including PowerPoint and PDF directly from your notebook's sources.
The Access Question Nobody Is Being Clear About
Here is the part worth understanding before you get excited about any of this for your classroom.
Coverage of this update describes it as available, at least initially, to Google AI Ultra subscribers and Workspace business customers with specific AI access tiers, rather than universally to every free NotebookLM user. This is a meaningfully different rollout pattern than the automatic Drive syncing update from a couple weeks earlier, which reached all Workspace customers and personal account holders without a paid tier requirement.
I want to be honest that the exact current access boundary for a standard teacher account, whether on a personal Google account or a school issued Workspace for Education account, was not something I could confirm with full certainty from available reporting at the time of writing. Some of the underlying model improvement likely benefits all users to some degree, since the core model gets upgraded regardless of which advanced features sit behind a paid tier. The specific agentic source discovery and cloud computer features appear more likely to be gated to paid tiers, at least in this initial rollout phase.
Before assuming any of the new capabilities described above are available to you, open your own NotebookLM account and check directly, or ask your school's IT department whether your Workspace for Education configuration includes AI Ultra access. Do not build a lesson or unit plan around a feature you have not personally confirmed you can access.
What This Means for the Unit Notebook Workflow Covered Elsewhere on This Site
If your NotebookLM access has not changed, meaning you are still on a standard free or standard Workspace for Education account, the practical day to day experience of building a unit notebook with curated sources and querying it should feel similar to before, with whatever quality improvement comes from the underlying model upgrade benefiting your existing workflow without you needing to do anything differently.
If you do have access to the new source discovery feature, there is a genuine classroom application worth considering carefully rather than adopting immediately. The feature that helps build a source list from a conversation is useful for a teacher's own unit preparation, where you already have professional judgment about what counts as a reliable source. It is a different question entirely whether you would want students using that same discovery feature directly, since it means the AI is choosing sources on the student's behalf using its own search capability rather than students working exclusively within a curated set you selected. The entire value of the unit notebook approach covered previously on this site rests on students working within sources you have vetted. A feature that lets the AI find its own sources during a student session works against that specific design principle unless you are deliberately teaching source evaluation as the skill itself.
For teacher side unit preparation, source discovery could genuinely save setup time. For a student facing research notebook, keep curating your own sources rather than turning discovery on, at least until you have thought through how it changes what your students are actually practicing.
Should You Change Anything About How You Use NotebookLM Right Now
Not urgently, and not until you have confirmed what you actually have access to. Check your account. If the new features are not visible yet, nothing about your existing workflow needs to change, and the underlying model quality improvement will apply to your notebooks regardless. If the new features are visible, use the source discovery tool for your own planning if it saves you time, and think carefully before extending the same discovery capability to a student facing notebook where the entire point was a bounded, teacher curated set of sources.
Written by

Muthu kumar
AI Education ReviewerI teach literacy and work across subjects with middle and high school students, and after three years in the classroom I have a pretty clear sense of what works and what just sounds good in a product demo. I started reviewing AI tools on TeachWithAI Tools because I wanted a space where teachers could get honest opinions without having to wonder if a recommendation was paid for. When I cover something outside my subject area I bring in someone who actually teaches it, because I think that matters. No sponsorships, no affiliate links. Just what I would genuinely tell a colleague in the staffroom.
Keep Reading


