# US 12,602,388 B2 · Generative search engine text documents > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 12,602,388 B2 - Document ID: US12602388B2 - Status: Granted - Kind: B2 - Application number: 18/643,806 - Filed: 2024-04-23 - Published: 2025-10-23 - Granted: 2026-04-14 - Earliest claimed: 2024-04-23 - Assignees: Microsoft Technology Licensing, LLC - Applicants: Microsoft Technology Licensing, LLC - Inventors: Mohamed Salman Ismail Gadit; Shivani Sridhar; Beverly Roberta Shen; Yu-An Lin; Ronak Digant Upadhyaya; Sriram Selvam; Avishek Dan - Web page: https://patentgrep.com/patent/US12602388B2 ## CPC - G06F 16/24578: {using ranking} - G06F 16/215: Improving data quality; Data cleansing, e.g. de-duplication, removing invalid entries or correcting typographical errors - G06F 16/9538: Presentation of query results - G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N) ## Abstract This disclosure describes utilizing a generative document system to dynamically build and provide generative text documents using one or more generative artificial intelligence (AI) models. For example, the generative document system efficiently utilizes various systems and one or more generative AI models to determine intents and topics, curate topic sections, and generate a generative text document that includes a directed answer along with select curated topic sections for search queries. In various implementations, the generative document system performs additional actions that enhance the efficiency and accuracy of operations used to produce generative text documents. Additionally, in many cases, these generative text documents provide a foundation for providing an interactive, intuitive, wide-ranging, and flexible curation of answers to users that address the corresponding search queries. ## Claims ### Claim 1 A computer-implemented method for generating one or more search result documents using one or more artificial intelligence (AI) models, comprising: obtaining a set of search result links from a search grounding system corresponding to a search query received from a client device; based on providing a search intent prompt and the set of search result links to a generative AI model, receiving a set of topic sections from the generative AI model based on the set of search result links; for a first topic section of the set of topic sections: obtaining a set of first topic section links from the search grounding system; and based on providing a first topic section prompt that includes the first topic section and the set of first topic section links to the generative AI model, receiving a first curated topic section with text and linked citations; and based on providing the first curated topic section and a second curated topic section with a generative document prompt to the generative AI model, receiving a generative text document that includes a directed answer followed by one or more curated topic sections. ### Claim 2 (depends on 1) The computer-implemented method of claim 1, wherein the search intent prompt includes instructions for the generative AI model to: determine a ranked list of intents for the search query based on the set of search result links; determine the set of topic sections from the ranked list of intents; and generate a search intent sentence that includes topic sections in the set of topic sections according to a topic section rank. ### Claim 3 (depends on 2) The computer-implemented method of claim 2, wherein: the search intent prompt further includes instructions for the generative AI model to generate the set of topic sections as a list; and each topic section in the list includes a topic title and a topic query. ### Claim 4 (depends on 3) The computer-implemented method of claim 3, wherein each topic section in the list also includes a set of alternative topic queries. ### Claim 5 (depends on 1) The computer-implemented method of claim 1, further comprising: obtaining multiple sets of topic section links for multiple topic sections from the search grounding system concurrently; and providing multiple topic section prompts for the multiple topic sections to one or more generative AI models concurrently. ### Claim 6 (depends on 1) The computer-implemented method of claim 1, wherein the first topic section prompt includes a first topic title, filtered content for the first topic section, the search query, a first topic query, and a search intent sentence for the search query. ### Claim 7 (depends on 1) The computer-implemented method of claim 1, further comprising: aggregating multiple curated topic sections received from the generative AI model into a combined topic section document; and deduplicating content within the combined topic section document based on the linked citations. ### Claim 8 (depends on 7) The computer-implemented method of claim 7, wherein the generative document prompt includes the combined topic section document, the search query, a first topic query, a search intent sentence for the search query, and instructions for the generative AI model to: generate the directed answer that directly addresses the search query; and generate the one or more curated topic sections from the combined topic section document. ### Claim 9 (depends on 8) The computer-implemented method of claim 8, wherein the generative AI model organizes the one or more curated topic sections in the generative text document based on an order of topic sections in the search intent sentence. ### Claim 10 (depends on 1) The computer-implemented method of claim 1, wherein the generative text document is an unformatted generative text document that includes directed answer text followed by one or more curated topic section texts. ### Claim 11 (depends on 1) The computer-implemented method of claim 1, further comprising generating a set of answer cards that corresponds to content within the generative text document. ### Claim 12 A system comprising: a processing system; and a computer memory comprising instructions that, when executed by the processing system, cause the system to perform operations of: obtaining a set of search result links from a search grounding system corresponding to a search query received from a client device; based on providing a search intent prompt and the set of search result links to a first generative AI model, receiving a set of topic sections from the first generative AI model based on the set of search result links; for a first topic section of the set of topic sections: obtaining a set of first topic section links from the search grounding system; and based on providing a first topic section prompt that includes the first topic section and the set of first topic section links to a second generative AI model, receiving a first curated topic section with text and linked citations; and based on providing the first curated topic section and a second curated topic section with a generative document prompt to a third generative AI model, receiving a generative text document that includes a directed answer followed by one or more curated topic sections. ### Claim 13 (depends on 12) The system of claim 12, wherein the first generative AI model, the second generative AI model, and the third generative AI model are different generative AI models. ### Claim 14 (depends on 12) The system of claim 12, wherein the operations further comprise, for the first topic section, providing the search query and a first topic query to the search grounding system to obtain a set of first topic section content that includes a set of topic section links, a set of linked content, and a set of answers each corresponding to the first topic section. ### Claim 15 (depends on 14) The system of claim 14, wherein the operations further comprise, for the first topic section, filtering the set of first topic section content based on a data limit threshold associated with the third generative AI model. ### Claim 16 (depends on 15) The system of claim 15, wherein the operations further comprise providing an alternative first topic query to the search grounding system with the search query and the first topic query, wherein: the set of first topic section content is separated into ranked lists according to the search query, the first topic query, and the alternative first topic query; and filtering the set of first topic section content includes: selecting a highest-ranked content from each of the ranked lists; and continuing to select a next highest-ranked content from each of the ranked lists until the data limit threshold is satisfied. ### Claim 17 A computer-implemented method for generating one or more search result documents using one or more artificial intelligence (AI) models, comprising: obtaining a set of search result links from a search grounding system corresponding to a search query received from a client device; based on providing a search intent prompt that includes the search query, the set of search result links, and query intent instructions to a generative AI model, receiving a set of topic sections from the generative AI model determined from a ranked list of intents corresponding to the search query and the set of search result links, wherein each topic section in the set of topic sections includes a topic title and a topic query; for each topic section in the set of topic sections: receiving a set of topic section links and corresponding content in response to providing the search query and the topic query to the search grounding system; and receiving a curated topic section with text having linked citations from the generative AI model based on providing a topic section prompt that includes the topic title, the set of topic section links, and the corresponding content to the generative AI model; and receiving a generative text document that includes a directed answer followed by one or more multiple curated topic sections from the generative AI model based on providing the generative AI model with a generative document prompt that includes the search query, the ranked list of intents, multiple curated topic sections, and a directed answer instruction. ### Claim 18 (depends on 17) The computer-implemented method of claim 17, wherein the search intent prompt does not include user information. ### Claim 19 (depends on 17) The computer-implemented method of claim 17, wherein the search intent prompt includes instructions for the generative AI model to: determine a ranked list of intents for the search query based on the set of search result links; determine the set of topic sections from the ranked list of intents; and generate a search intent sentence that includes topic sections in the set of topic sections according to a topic section rank. ### Claim 20 (depends on 17) The computer-implemented method of claim 17, wherein the generative text document is an unformatted generative text document that includes directed answer text followed by one or more curated topic section texts.