Google’s Gemini 2.5 interacts via browser, not API
On October 7–8, 2025 (covered in recent AI news), Google introduced its Gemini 2.5 Computer Use model, allowing AI agents to interact with web pages via a browser interface (e.g. filling forms, clicking), rather than backend APIs. It supports 13 actions and shows strong performance on web and UI tasks. For developers, this opens doors to automating workflows even when no APIs exist. It marks a shift toward “agentic” AI bridging UI and reasoning layers.