Google I/O 2026: Project Genie Will Now Let You Explore Street View Locations With a ‘Creative Twist’

At Google I/O 2026, the Mountain View-based tech giant announced a new capability for Project Genie. The new experience connects the cutting-edge large language model (LLM) with Google's Street View to let users explore different locations on the map with a “creative twist.” The feature will be available to those on the highest subscription tier of Gemini as the company also starts rolling out Project Genie to more users. Apart from a fun, interactive experience, the tech giant also pitches the Street View grounding as a tool to help train artificial intelligence (AI) agents and embedded robots.

Google Releases New Project Genie Experience

In a blog post, the tech giant announced and detailed the new Project Genie experience. Notably, ever since previewing the Genie 3 AI model last year, Google has been working on new experiences built on its capabilities. For those out of the loop, Genie 3 is a world model, meaning that it is a general-purpose model that specialises in generating interactive, 3D worlds. The company released a version of the model to the Ultra subscribers in 2025.

Now, Google is rolling out Project Genie, including the new Street View capability, to all adult AI Ultra $200 (roughly Rs. 19,356) subscribers globally. Eligible users should remember that Project Genie is an experimental research prototype in Google Labs.

Coming to the experience itself, it can be understood as interactive worlds based on street views, with fun geographical filters. For instance, a user can select the Golden Gate Bridge and then add the underwater filter to explore the region as a diver, as the scenery gets filled with seaweed and aquatic life forms. The best part is that the entire AI-generated world is interactive.

Google says the feature connects Project Genie with Street View, as the latter acts as the grounding tool. This means the world generation is more accurate and based on the data from Google Maps' Street View. Once it is created, the user will see an avatar of themselves, and using W, A, S, and D keys on the keyboard, can navigate the environment.

The company highlighted that this expansion can also be used to provide virtual environments for AI agents or embedded robots to navigate and interact with the complexities of the real world.