# US 2026/0228295 A1 · Website Generation Using Language Models > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 2026/0228295 A1 - Document ID: US20260228295A1 - Status: Published application - Kind: A1 - Application number: 19/042,542 - Filed: 2025-01-31 - Published: 2026-08-06 - Earliest claimed: 2025-01-31 - Assignees: HOSTINGER operations, UAB - Applicants: HOSTINGER operations, UAB - Inventors: Andrius Grybauskas; Arminas Maciulskas; Tomas Rasymas - Web page: https://patentgrep.com/patent/US20260228295A1 ## CPC - G06F 16/958: Organisation or management of web site content, e.g. publishing, maintaining pages or automatic linking - G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N) ## Abstract Generating a website includes selecting website sections from a predefined section database based on a determined website type; generating, using a language model, keywords for image search by analyzing a website description; converting the keywords into vector embeddings; performing, based on the vector embeddings, a vector search in an image database to identify relevant images; ranking the relevant images based on similarity scoring between vector embeddings of the website description and embeddings of image textual descriptions; assembling the selected website sections and one or more of the ranked images into a hierarchical data structure; and rendering the hierarchical data structure in a web browser. ## Claims ### Claim at position 1 1-20. (canceled) ### Claim 21 A method for generating a website, comprising: selecting website sections from a predefined section database based on a determined website type; generating, using a language model, keywords for image search by analyzing a website description; converting the keywords into vector embeddings; performing, based on the vector embeddings, a vector search in an image database to identify relevant images; ranking the relevant images based on similarity scoring between vector embeddings of the website description and embeddings of image textual descriptions; assembling the selected website sections and one or more of the ranked images into a hierarchical data structure; and rendering the hierarchical data structure in a web browser. ### Claim 22 (depends on 21) The method of claim 21, further comprising: generating an image using an image generation model in response to the vector search not identifying any relevant image having a similarity score above a predefined threshold. ### Claim 23 (depends on 21) The method of claim 21, wherein the image database comprises image embeddings that each include: a first embedding representing a textual description of the image; and a second embedding representing visual features of the image. ### Claim 24 (depends on 21) The method of claim 21, wherein selecting website sections comprises: determining, using the language model, a website type based on the website description; and selecting sections from the section database based on the determined website type. ### Claim 25 (depends on 21) The method of claim 21, further comprising: generating, using the language model, search engine optimization metadata for at least one page based on the website description and user preferences. ### Claim 26 (depends on 21) The method of claim 21, wherein ranking the relevant images comprises: computing a similarity score for each image based on both: semantic similarity between the website description and the image textual description; and visual coherence with the selected website sections. ### Claim 27 (depends on 21) The method of claim 21, further comprising: analyzing the website description using the language model to extract explicit preferences for at least one of colors, fonts, or layout style. ### Claim 28 A system for generating a website, comprising: a memory subsystem; and processing circuitry, the processing circuitry configured to execute instructions stored in the memory subsystem to: select website sections from a predefined section database based on a determined website type; generate, using a language model, keywords for image search by analyzing a website description; convert the keywords into vector embeddings; perform, based on the vector embeddings, a vector search in an image database to identify relevant images; rank the relevant images based on similarity scoring between vector embeddings of the website description and embeddings of image textual descriptions; assemble the selected website sections and one or more of the ranked images into a hierarchical data structure; and render the hierarchical data structure in a web browser. ### Claim 29 (depends on 28) The system of claim 28, the processing circuitry further configured to execute instructions stored in the memory subsystem to: apply a color palette and typography to the website, wherein the color palette is selected based on user preferences or industry standards, and wherein the typography is mapped to a font style consistent with the website type. ### Claim 30 (depends on 28) The system of claim 28, wherein, to assemble the selected website sections and ranked images, the processing circuitry configured to execute instructions stored in the memory subsystem to: insert the images into sections based on contextual relevance; and adjust element layout parameters to maintain compatibility with responsive design characteristics. ### Claim 31 (depends on 28) The system of claim 28, the processing circuitry further configured to execute instructions stored in the memory subsystem to: present a preview of the rendered website via a web interface as the hierarchical data structure is updated. ### Claim 32 (depends on 28) The system of claim 28, wherein, to render the hierarchical data structure, the processing circuitry configured to execute instructions stored in the memory subsystem to: generate a Document Object Model (DOM) based on the hierarchical data structure; and display the DOM in a browser-compatible format. ### Claim 33 (depends on 28) The system of claim 28, wherein the hierarchical data structure comprises page objects representing pages of the website, section objects associated with the page objects, and element objects within each section object, each element object including styling and positioning metadata. ### Claim 34 (depends on 28) The system of claim 28, wherein, to perform the vector search, the processing circuitry configured to execute instructions stored in the memory subsystem to: compute a vector similarity between the vector embeddings of the keywords and the image embeddings using a vector similarity search retrieval algorithm. ### Claim 35 (depends on 28) The system of claim 28, the processing circuitry further configured to execute instructions stored in the memory subsystem to: filter the identified images based on one or more design constraints selected from image shape or dimensional characteristics. ### Claim 36 One or more non-transitory computer readable media storing instructions operable to cause one or more processors to perform operations for generating a website, the operations comprising: selecting website sections from a predefined section database based on a determined website type; generating, using a language model, keywords for image search by analyzing a website description; converting the keywords into vector embeddings; performing, based on the vector embeddings, a vector search in an image database to identify relevant images; ranking the relevant images based on similarity scoring between vector embeddings of the website description and embeddings of image textual descriptions; assembling the selected website sections and one or more of the ranked images into a hierarchical data structure; and rendering the hierarchical data structure in a web browser. ### Claim 37 (depends on 36) The one or more non-transitory computer readable media of claim 36, wherein generating the keywords comprises: analyzing the website description to isolate descriptive phrases; and mapping the descriptive phrases to keywords using few-shot prompt examples. ### Claim 38 (depends on 36) The one or more non-transitory computer readable media of claim 36, the operations further comprising: updating the hierarchical data structure in response to a user modification to an image or section, wherein the website is rendered in real-time based on the updated data structure. ### Claim 39 (depends on 36) The one or more non-transitory computer readable media of claim 36, the operations further comprising: generating an image using an image generation model in response to the vector search not identifying any relevant image having a similarity score above a predefined threshold. ### Claim 40 (depends on 36) The one or more non-transitory computer readable media of claim 36, wherein the image database comprises image embeddings that each include: a first embedding representing a textual description of the image; and a second embedding representing visual features of the image.