Google Shakes Up Its Browser Agent Team Amid OpenClaw Craze

Google is reorganizing the team behind Project Mariner, its research prototype for an AI agent that could navigate the Chrome browser to click, scroll, and fill out forms. According to WIRED, staff from Google Labs have been moved to higher-priority projects. A Google spokesperson confirmed the changes, stating the computer-use capabilities developed under Mariner will be incorporated into the company's broader agent strategy, including the recently launched Gemini Agent.

This strategic pivot comes as the AI industry's focus shifts dramatically from browser-based agents to more efficient command-line agents like OpenClaw and Claude Code. Nvidia CEO Jensen Huang recently called OpenClaw a 'new operating system for agentic computers,' highlighting its strategic importance. In contrast, standalone browser agents have seen limited adoption; Perplexity's Comet reached 2.8 million weekly active users in December 2025, while OpenAI's ChatGPT Agent reportedly fell below 1 million, a fraction of ChatGPT's user base.

Experts point to computational efficiency as a key differentiator. Kian Katanforoosh of Workera notes that command-line interaction is '10 to 100X less steps' for LLMs than processing screenshots of graphical interfaces. However, the need for GUI interaction persists. Ang Li, CEO of Simular, argues an '80/20 split' will remain, as legacy systems like healthcare websites often lack APIs for terminal agents. The evolution continues, with startups like Standard Intelligence developing video-trained models claimed to be 50X more efficient.