Listen: Google AI Search Update: Completely New Grounding Format
An observation of a new, custom grounding context format for Gemini that deviates from the traditional index-based model used in previous prompt types.
Transcript
Google Gemini is quietly changing how it handles grounding context, and the shift is breaking the old system. Previously, Gemini relied on a simple, predictable structure. It organized background search results using a basic model of numbered indexes, like index one, index two, and so on.
But recently, a completely new format has emerged. The grounding data now seems customized to the specific type of prompt you enter. For example, if you ask about a writer's latest articles, Gemini no longer just dumps basic list items. Instead, it groups the results with dedicated fields for the author's name, a collection of relevant snippets, the source title, and the link.
For other topics, like cycling gear or search engine optimization tools, the format adapts again, sometimes bundling multiple different snippets under a single index number.
This means Gemini is moving away from a one-size-fits-all structure to a highly dynamic, context-aware format. It is a fascinating shift in how the artificial intelligence structures web data, and developers are still digging in to understand exactly how these custom formats are triggered.