Google DeepMind is developing an AI-powered computer mouse cursor based on Gemini, aiming to rethink a technology that has remained unchanged for over half a century.
“We are exploring new AI-based possibilities that will help the cursor not only understand what it points to but also why it matters to the user,” the company blog states.
The project addresses a common issue: the standard AI tool operates in a separate window, requiring users to drag their content there.
“We aim for the opposite: an intuitive artificial intelligence that interacts with users across all their tools without interrupting the workflow,” the company declared.
Four Principles of the AI Cursor
Google DeepMind has developed four principles that shift the transmission of context and intentions from the user to the computer, replacing text prompts with more intuitive interaction methods.
Maintaining Workflow
AI functions should work across all applications, so the Google DeepMind cursor prototype is available wherever the user operates. For instance, one can hover over a PDF file and request a brief summary in list form to insert into an email.
Another example is hovering over a table with statistical data and asking to create a pie chart.
Show and Tell
Modern neural networks require precise instructions, and to get a quality response, users need to craft detailed prompts. The AI cursor simplifies this process: it captures visual and semantic context, allowing the model to “see” and understand what is important.
In Google DeepMind’s experimental system, it is sufficient to point to a paragraph, image fragment, or code block, and the LLM will understand what assistance is needed.
The Power of “This” and “That”
In everyday communication, people rarely use long and detailed phrases. They say, “Fix this,” “Move that here,” or “What does this mean,” while relying on gestures and overall context.
“An AI system capable of understanding this combination of context, gestures, and speech will allow users to formulate complex requests naturally and succinctly without using complex prompts,” the Google DeepMind blog notes.
Pixels for Interaction
For decades, computers have only tracked the direction of gaze. Artificial intelligence can understand what the user is pointing at and transform pixels into structural objects for instant interaction.
“A photo with sketches turns into an interactive to-do list, a still from a travel video becomes a link to book a table at that charming restaurant,” noted Google DeepMind.
Implementation
Google DeepMind is implementing these principles in Chrome and the new interface for Googlebook laptops.
“Starting today, instead of entering complex queries, you can use the cursor to ask Gemini in Chrome about the part of the web page that interests you,” the blog states.
For example, you can select several items on a page and ask to compare them or indicate where to visualize a new sofa in the living room.
Soon, the company will launch Magic Pointer in Googlebook. The feature will allow using Gemini with a single finger swipe.
Googlebook
Google has introduced a “new category of laptops” based on Gemini’s capabilities, combining Android and ChromeOS.
“Over 15 years ago, we introduced the Chromebook—a laptop designed for a cloud-first world. Today, as we transition from OS to an intelligent system, we see an opportunity to rethink the concept of laptops,” the announcement states.
Googlebooks are designed for Gemini Intelligence and the new AI cursor. The devices feature a “Create Widget” function with voice command support.
Gemini can search the internet or connect to Google apps like Gmail and Calendar to create a unified personalized dashboard.
The tool is built on the Android technology stack, ensuring seamless operation across multiple devices: the “Quick Access” feature allows viewing, searching, and inserting files from a phone to a laptop without manual transfer.
The first Googlebooks are being developed in collaboration with Acer, ASUS, Dell, HP, and Lenovo.
Gemini Intelligence
Gemini Intelligence is a suite of AI functions for Android devices. In the summer of 2026, they will first appear on Samsung Galaxy and Google Pixel, then on other gadgets.
“This system combines top-notch hardware and innovative software to help you stay ahead, proactively solving tasks throughout the day while ensuring data privacy and full control,” the announcement states.
Some agent capabilities of Gemini were already demonstrated in early 2026 on Samsung smartphones, such as ordering food or taxis. Soon, the assistant will be able to perform more complex tasks: simply photograph a tour brochure, and the bot will find similar options on Expedia.
In Chrome, the AI assistant will help find, summarize, and compare content across different pages. The Rambler feature on the Gboard keyboard will allow dictating text, then highlight important fragments and compile them into a coherent message without filler words and repetitions.
Back in April, Google introduced Gemma 4—a new family of open AI models for advanced reasoning and agent workflows.