Google is testing a new “Projects” feature for its Gemini AI, allowing users to integrate files like documents and code for contextual reference within conversations, potentially streamlining research tasks and workflow management. The precise rollout timeline remains unconfirmed, with initial availability possibly restricted to enterprise clients.
BREAKING 🚨: Google is working on Projects for Gemini! There, it will be able to run Research tasks across uploaded documents. pic.twitter.com/Yc6R06ALHu
— TestingCatalog News 🗞 (@testingcatalog) August 15, 2025
This “Projects” functionality enables users to add various files, documents, or personal code, which Gemini can then reference during ongoing conversations. Google states that users can “start by adding files to the project. Gemini can reference these files in project conversations and use them to generate new documents and code.” This feature facilitates the grouping of related tasks, establishing a dedicated workspace within Gemini.
Concurrently, Google is developing Gemini 3, anticipated to be its most advanced model. Google DeepMind, the team responsible for Gemini, recently introduced Genie 3. Genie 3 allows users to transform text prompts into interactive, playable virtual worlds.
The system can simulate real-time navigation within these generated environments, including the depiction of a human figure walking. Genie 3 renders these simulated worlds at 720p resolution and 24 frames per second, maintaining physical consistency for several minutes.
