Generative search results documents based on enhanced search results using generative artificial intelligence (AI) models
Topics: Microsoft
This Microsoft patent describes a system called the “Generative Document System” that automatically builds structured, AI-generated search results pages in response to user queries. Instead of simply returning a list of links, the system uses multiple small, specialized AI models working together and often simultaneously to analyze a search query, find the most relevant sources, and produce a complete document with a direct answer, topic sections, and cited sources.
This patent applies directly to Microsoft’s Bing search engine and its AI-powered search features (such as Bing Deep Search and Bing’s AI Mode). It is highly relevant to GEO (Generative Engine Optimization) because it reveals exactly how Microsoft’s AI search decides which queries get a full AI-generated response, how it selects sources to cite, and how content is structured and ranked within that AI-generated document. Understanding this process tells content creators and SEO professionals what kinds of content and signals increase the chances of being cited inside an AI-generated answer.
