Google has officially integrated “computer use” as a native, built-in tool within its Gemini 3.5 Flash model, a major milestone for enterprise automation and AI agent development. Previously limited to a standalone Gemini 2.5 model, this native inclusion enables Gemini 3.5 Flash to achieve optimal performance for complex, agentic tasks. Developers can now utilize this feature to build tailored, cross-platform agents capable of seeing, reasoning, and executing precise actions across mobile, browser, and desktop environments. This development significantly improves capabilities for enterprise workflows, including professional knowledge work and continuous software testing.
Also Read: Harmonizing Innovation: Meta’s AudioCraft Redefines the Future of Generative AI for Sound
To address security concerns like prompt injection risks in live corporate ecosystems, Google has integrated targeted adversarial training along with two crucial enterprise safeguards: requiring explicit user confirmation for sensitive or irreversible actions and automatically terminating tasks if indirect prompt injections are spotted. Available today via the Gemini API and the Gemini Enterprise Agent Platform, early adopters are already realizing its immense value. Highlighting its practical impact, Alvin Stanescu, Senior Director at UiPath, noted that the technology represents a leap forward, stating, “with Gemini 3.5 Flash’s computer use capabilities, we are unlocking unprecedented speed and cost-efficiencies for UI-based automation tasks.” Ultimately, this native upgrade positions Gemini 3.5 Flash as a highly versatile and scalable tool for secure, long-horizon B2B digital transformation.
