Watch: Content Node
A single element in Chrome's page representation. Each node carries a type (one of 21, from headings and links to form controls and dialogs), bounding box coordinates, text styling, accessibility metadata, and an ID that the AI uses to reference or interact with that specific element.
Transcript
When Google Chrome prepares a web page for the Gemini artificial intelligence, it translates the page's visual layout into a structured hierarchical tree. Every visible element on the page becomes a single content node.
Chrome assigns each node one of twenty-one distinct types. These types tell the AI how to interact with the element. For example, a heading signals the structure of the document, an anchor is something the AI can click, and a form control is an input it can fill out.
But these nodes carry more than just a type. They also contain precise geometric data, including the element's outer and visible boundaries in screen pixels. They hold text styling details like font size and color, along with crucial accessibility metadata, such as whether an element is focusable or disabled.
To keep things efficient, Chrome filters out the background noise. It strips away structural containers and layout wrappers that don't add meaning. On a typical webpage, Chrome might look at nearly five hundred nodes in the rendering tree, but it will only pass about two hundred of those tagged nodes to Gemini. This distilled, structured format gives the AI exactly what it needs to read, understand, and navigate the web just like a human would.