Flipbook: The AI-Powered “Visual Browser” That Turns the Entire Internet into Generated Images

Just when we thought we’d seen the limits of generative AI — from fully AI-made videos to entire video-game interfaces created on the fly — the next frontier has arrived: the internet itself.

Flipbook: The AI-Powered “Visual Browser” That Turns the Entire Internet into Generated Images A small team of former researchers and engineers from OpenAI, Humane, and Apple has launched Flipbook, an experimental browser that reimagines web browsing as an infinite sequence of AI-generated images. There is no HTML. No hyperlinks. No text layers or traditional UI elements.

Every “page” you see is a single, high-resolution image painted in real time by a large multimodal model. Click anywhere on that image — a person, an object, a word rendered as pixels — and the system instantly generates a brand-new image that dives deeper into whatever caught your eye.

The concept is disarmingly simple and radical at the same time. As the creators put it: “A picture is worth a thousand words, yet we fill our screens with mostly text and colored rectangles.”

Flipbook does away with the rigid scaffolding of the traditional web. The entire experience is just pixels — beautiful, rich, and generated on demand.

Flipbook: The AI-Powered “Visual Browser” That Turns the Entire Internet into Generated Images Behind the scenes, content is assembled through a clever hybrid approach: an “agentic” web-search system pulls up-to-date information, which is then combined with the image model’s own internal knowledge.

The result, according to the team, achieves factual accuracy on par with ChatGPT, Gemini, or Claude. Even the text you read on each page isn’t real text — it’s pixels drawn by the same model, which occasionally produces charmingly imperfect lettering or slightly misplaced words.

For those who want even more immersion, there’s an experimental “live video stream” mode. Toggle it on and the static images turn into a continuous 1080p video feed. A custom video model animates smooth transitions between pages while the image generator keeps painting the next visual. It’s resource-heavy, a little unpredictable, and currently runs on separate systems (with plans to unify them), but the effect is mesmerizing — like flipping through an infinite, living picture book.

The project is the work of Zain Shah (ex-OpenAI), Eddie Jiao (ex-Humane), and Drew Carr (ex-Apple). Backed by South Park Commons and with inference sponsored by Modal, Flipbook launched quietly but has already drawn huge interest — so much that new visitors are currently funneled through a wait queue to keep the generation servers from melting.

Of course, the creators’ claim of “real time” is… optimistic. In practice, each new page takes several seconds to appear — a delay that instantly transports veteran users back to the golden (or should we say dial-up) era of 56k modems. It’s nowhere near the instant responsiveness we expect from modern browsers. Yet the experience remains strangely compelling. There’s something hypnotic about watching a fresh, richly illustrated page materialize exactly tailored to your curiosity.

Flipbook: The AI-Powered “Visual Browser” That Turns the Entire Internet into Generated Images Flipbook is still a prototype, explicitly labeled as experimental. The team openly acknowledges current limitations: occasional hallucinations, imperfect text rendering, and the fact that it’s not yet ready for complex actions like filling forms or persistent data storage.

But they’re already thinking ahead — toward faster, more accurate models that could one day support real interactivity, memory across sessions, and even replace entire apps.

In an age when AI is already generating videos, games, and now entire web pages, Flipbook feels like a logical — and slightly rebellious — next step.

It’s less a replacement for Chrome or Safari and more a provocative proof-of-concept: what if the web didn’t have to be built out of code and rectangles anymore? What if it could simply be summoned, image by image, exactly when and how we need it?

Right now it’s slow, a little glitchy, and occasionally hallucinatory.

But it’s also undeniably cool — the kind of wild experiment that makes you wonder whether, a few years from now, we’ll look back at today’s static websites the same way we now view Geocities pages from the ’90s.

If you want to try it yourself, head to flipbook.page — just be prepared for a short wait in line. The future of browsing, it seems, is being drawn one pixel at a time.