Agent S2

Introducing the Gemini 2.5 Computer Use model, a specialized agent model built on top of Gemini 2.5 Pro’s visual reasoning capabilities, designed to interact directly with user interfaces (UIs). It is exposed via a new computer-use tool in the Gemini API, with inputs that include the user’s request, a screenshot of the UI environment, and a history of recent actions. The model generates function calls corresponding to UI actions like clicking, typing, or selecting, and may request user confirmation for higher-risk tasks. After each action is executed, a new screenshot and URL are fed back into the model to continue the loop until the task completes or is halted. It is optimized primarily for web browser control and shows promise for mobile UI interaction, though it is not yet suited for desktop OS-level control. In benchmarks across web and mobile control tasks, Gemini 2.5 Computer Use outperforms leading alternatives, delivering high accuracy at lower latency.

The Agentic Era marks a transformative shift from traditional application-centric computing to a new frontier defined by agentic AI, autonomous, context-aware systems capable of acting, learning, and collaborating within complex, dynamic environments. These intelligent agents don’t just respond to commands; they perform complete tasks, retain memory and context via large language models tailored for specific domains, and can scale across industries into the tens of millions. This evolution brings the need for a new operational mindset, AgenticOps, and a reimagined management interface built around three guiding principles, keeping humans thoughtfully in the loop to provide creativity and judgment, enabling agents to operate across siloed systems with cross-domain context, and deploying purpose-built models fine-tuned for their distinct tasks. Cisco brings this to life through AI Canvas, the industry’s first generative, shared workspace driven by a multi-data, multi-agent architecture.