Google has reorganized the team behind Project Mariner, its AI agent designed to navigate Chrome and complete tasks on a user’s behalf, Wired reported on March 19. Several Google Labs staffers who worked on the research prototype have moved to higher-priority projects, according to two people familiar with the matter.
A Google spokesperson confirmed the changes but said the computer-use capabilities developed under Project Mariner will be incorporated into the company’s broader agent strategy, including the recently launched Gemini Agent.
Browser Agents Are Struggling
The shakeup reflects a broader industry problem: browser agents, once positioned as the next major consumer AI product, have not found their audience.
Wired cites specific adoption numbers. Perplexity’s Comet browser agent reached 2.8 million weekly active users in December 2025. OpenAI’s ChatGPT Agent reportedly fell to less than 1 million weekly active users in recent months. Compared to the hundreds of millions of users talking to ChatGPT weekly, browser agent usage is a rounding error.
The products can click, scroll, and fill out forms on a webpage, mimicking human interaction. The problem is reliability. Browser-based automation is fragile — websites change layouts, CAPTCHAs block automated access, and multi-step workflows break at unpredictable points.
Command-Line Agents Took a Different Path
OpenClaw and Claude Code control computers through the command line, which has proven more reliable for completing tasks. These systems interact with file systems, APIs, and shell commands directly, bypassing the visual rendering layer that makes browser automation brittle.
OpenClaw’s browser-use capability exists as one feature among many, rather than as the entire product. That approach has resonated: Nvidia CEO Jensen Huang compared OpenClaw to a new operating system at GTC 2026, saying “Every company in the world today needs to have an OpenClaw strategy.”
Google CEO Sundar Pichai highlighted Project Mariner at Google I/O 2025 as a flagship demonstration of Google’s agent capabilities. One year later, the team behind it is being reassigned.
What Google Does Next
Google folded some of Project Mariner’s computer-use capabilities into the Gemini Agent, which shipped with Gemini 3. The spokesperson’s framing — that the technology lives on even if the team is reorganized — is standard for projects that get absorbed into larger efforts.
The question is whether Google can accelerate its agentic AI roadmap fast enough. OpenClaw’s acquisition by OpenAI gives it resources and distribution. Nvidia’s NemoClaw provides enterprise infrastructure. Anthropic’s Claude Code has carved out the developer market. Google’s response so far has been to rearrange its org chart.
Google I/O 2026 would be the natural venue for an accelerated agent announcement. Whether the Mariner team’s work shows up there as a product, rather than a demo, will determine if this shakeup was a course correction or an admission that Google fell behind.