Google has quietly pushed one of the most consequential updates to its research assistant this year: Gemini Notebook can now transform the notes, documents and sources stored inside a notebook into a fully interactive website — a navigable periodic table, a historical timeline, a sales dashboard — without the user writing a single line of code. The capability began rolling out globally on the web on 16 November 2026 for users aged 18 and over, including Google Workspace subscribers, according to Google’s official blog.
- Prompt-driven creation: Users type “make me a site” in a notebook chat alongside a description of what they need, then open the result in a new tab or share it.
- Three headline use cases: Visual study aids such as interactive periodic tables and timelines, data dashboards for exploring business figures, and project trackers with tasks and deadlines.
- Web first, mobile later: The feature is live on the web globally; Google says a mobile app version is coming in the weeks ahead.
From static sources to something you can click
The mechanics are deliberately simple. Inside an existing notebook stocked with material, the user issues the instruction “make me a site” together with a precise description of the desired output. Gemini Notebook then analyses the sources and generates a page that can be manipulated rather than merely read. Once generated, the site is stored within the notebook itself and appears in the Studio area, from where it can be opened in a new tab or shared via a link.
Google’s own demonstrations illustrate the range. In one, a notebook packed with chemistry material becomes a site titled Understanding the Periodic Table, complete with clickable elements, distinct sections and a five-question quiz at the end to test retention. In another, sales documents are converted into a control panel with region filters, charts comparing actual revenue against assigned quota, and visualisations of the sales pipeline. Everything stays inside the Gemini Notebook interface.
The company frames the feature around three broad categories: academic support through visual explainers, business intelligence through interactive data panels, and project management through task and deadline trackers. The underlying logic is that research material should stop being static — a syllabus, a report or a dataset becomes a small application the user can operate.
What Google has not disclosed
The announcement is notably light on technical detail. Google has not named the model powering the generation, nor listed supported languages, nor specified usage limits, nor given the feature a formal product name beyond “interactive sites”. The blog post was authored by Maria Ines Aranguren, a software engineer at Google Labs, rather than by an executive — a low-key profile that fits what this is: a lab feature being tested at scale.
There is also no announced additional cost. The capability is included in whatever access a user already has to the platform, which makes the sharing dimension particularly interesting for teams: generated sites function as lightweight consultation apps that anyone can open in a browser without installing anything.
“Gemini Notebook can now turn your sources into interactive sites like dashboards and visual explainers. To get started, simply type ‘make me a site’ in your Gemini Notebook chat, along with a description of what you want to create, then select ‘Open in a new tab’ to view or share it.”
Early hands-on accounts suggest generation takes a few minutes depending on the volume of sources in the notebook, and that the output design is polished and minimalist. It is worth noting that creating a web page from notebook content was not previously impossible — users could add a notebook as a source in the Gemini app and request a site through Canvas. What is new is that the whole workflow now happens without leaving Gemini Notebook, with the result stored in the Studio section.
The upside, and the caveats worth keeping in mind
The strategic significance goes beyond convenience. By building the site strictly on sources the user has supplied, the feature reduces — though does not eliminate — the risk of the model inventing data. That anchoring has been a consistent theme in Google’s recent work on the platform, which has already moved from summarising sources to generating podcast-style audio overviews and interactive study guides.
Two caveats deserve attention. First, a generated site can contain design or logic errors that slip past unnoticed if nobody reviews it; looking like an app is not the same as working like one. Second, there is the dependency question: the easier generation becomes, the less users understand what sits underneath. For learning, explaining and exploring one’s own data, however, the direction is clearly right — and the boundary between reading about something and using a tool built on that something is now considerably blurrier.