Former OpenAI Researcher Releases Flipbook Prototype: Skip HTML, Generate Every Pixel Directly with AI Video Models

According to Omdena Beating monitoring, former OpenAI researcher Zain Shah and the team released Flipbook, an experimental prototype that uses an AI model to directly generate screen pixels, replacing traditional web technologies such as HTML and CSS. Each "page" seen by users is an image generated by AI, and clicking on any area of the image generates a new image to delve deeper. The entire interface has no HTML code, no fixed links, no predefined buttons, and even the text is pixelated within the image.

The video mode is based on the open-source DiT (Diffusion Transformer) video generation model LTX Studio from the Israeli company Lightricks, optimized to stream in real-time at 1080p 24fps to users' screens via WebSocket, with backend support from Modal Labs' serverless GPU. Shah stated that currently Flipbook has limited functionality, designed around visual exploration, but it demonstrates a larger direction: as models become more accurate and stateful, the future potential expansion to structured UI, including programming scenarios.

Shah previously worked on AI and robotics at OpenAI, then served as a Creative Technologist at Samsung and is also a YC S13 alum. Team members also include former Humane and Slack engineer Eddie Jiao, and former Apple engineer Drew O'Carr.