Meet uKit AI uKit AI is a technology stack we develop based on machine learning and generative design. The aim of the project is to automatically manage the content of a website its layout as well as its look and feel using the estimates and actual data of our survey about the things visitors focus on a web page their individual preferences and goals they pursue when visiting a website. uKit AI is able to manage the content of any website whether created with the uKit platform or any other service. As an example, we invite you to review a typical website First, let’s learn more about it. Because this website was not built with uKit uKit AI starts its work with importing the content published on this site. Now you can see the prototype of the uKit AI interactive visualization system which will allow our users to watch their sites being automatically rebuilt and will also engage them into a visual dialog with AI in terms of all the content that needs to be processed. This website contains about 50 pages its layout and design are relatively simple We use the WebGL technology for uKit AI to conduct visualizations which lets us visualize all pages of a typical site at the same time and hold a visual dialog with users about the entire content of their websites. During the first step, uKit AI analyzes the low-level DOM structure of the site and using machine vision identifies individual widgets that the website is made of. Then the system separates the common design elements that can be found on each page from the unique content. The information we’ve received this way is then transformed into a content tree a hierarchical structure reflecting the original way the information is arranged. If a user wants just to import the information uKit AI can transfer the content page by page keeping the same structure the imported website had. However, the real purpose of uKit AI is generative redesign in other words, the automatic synthesis of the website’s new design. And effective redesign most often requires rearranging the layout to a significant degree. In this case, the original content tree rebuilds itself so that the key info that should get the maximum exposure shows up in the root of the content tree. Next, the system groups information blocks into pages and adds static design elements. After that, the system performs a primary heuristic evaluation of the overall aesthetics of the obtained design version and predicts the distribution of users’ attention in the form of heat maps. As part of the generative design concept, to ensure the best result, the system produces several layout options and then selects the best one. In the end, the final version becomes available for editing in the uKit builder. Possibilities of uKit AI generative redesign coupled with machine learning from real traffic will serve as a basis for uKit AI 2.0. The project that goes for an ICO. uKit AI 2.0 is aimed at automatic generation and optimization of landing pages constantly training itself on the real flow of visitors and ensuring the continuous improvement of websites.