Google has unveiled Project Genie, a new AI-powered web application that allows users to create and explore interactive digital worlds. The announcement was made by Google CEO Sundar Pichai, who said he has personally experimented with the tool and described it as “out of this world.” Designed as a prototype, the web app enables users to turn ideas into immersive environments that can be navigated rather than just viewed. Google says the release is experimental, aimed at understanding how people interact with AI-generated worlds and what future use cases may emerge from this technology.
What is Project Genie?Project Genie is an experimental web-based AI tool developed by Google that lets users design their own interactive virtual worlds. Unlike traditional image or video generators, it creates environments that users can actively explore. By using simple text prompts or images, users can build scenes, characters and settings that respond dynamically as they move through them. The focus of Project Genie is on creativity and exploration, allowing people to experience AI-generated spaces in a more immersive way. Google positions it as a step toward more advanced simulations that could eventually bridge digital and real-world applications.
Technology behind Project GenieProject Genie is powered by Genie 3, Google’s AI world-generation model that was introduced last year as a limited research preview. The tool also integrates Nano Banana Pro and Gemini to enhance visual quality and responsiveness. According to Sundar Pichai, this combination of models enables the app to generate interactive environments instead of static visuals. To demonstrate its capabilities, Pichai shared a clip of an astronaut exploring a space station in Earth’s orbit created using Project Genie, highlighting how the technology brings depth, motion and realism to AI-generated worlds.
How Project Genie worksThe creation process in Project Genie begins with a step called “world sketching.” Users provide text descriptions or images, which the AI uses to generate complete environments or characters. These images can include real-world objects, such as toys, that are transformed into virtual elements. Once the world is created, users can explore it from different perspectives, including first-person and third-person views. As users move, the AI generates the surroundings in real time, simulating physics and interactions to make the experience feel more natural and engaging.
Features, limits and availabilityProject Genie also allows users to remix existing worlds available on the platform by modifying them with custom prompts. However, there are limitations. Each generated world can only be explored for 60 seconds, and Google notes that visuals may not always look fully realistic. Users may also experience higher latency while controlling characters. Currently, Project Genie is available only to Google AI Ultra subscribers in the US. Google DeepMind says the experimental release will help identify new use cases, including the potential for training robots to navigate real-world environments in the future.