Google reorganizes its browser agent team while computer use agents like OpenClaw gain traction in the AI space.
Silicon Valley -: Google is changing its browser agent team while AI tools compete for users.
Project Mariner helps people browse the web using AI. The team working on it has changed roles. Some people moved to other important projects at Google.
Google says they will use what they learned from Project Mariner in new ways. They already put some of these abilities into their Gemini Agent product. This change happened after new tools like OpenClaw became popular.
OpenClaw uses commands instead of clicking on web pages. Many AI labs want to make agents that help people do tasks. These agents could control computers and help businesses soon.
Big company leaders think every business needs plans for OpenClaw-like tools. Browser agents can click and scroll like humans. But not many people use them compared to talking AI assistants.
Perplexity’s Comet browser tool had 2.8 million weekly users in December. OpenAI’s ChatGPT Agent had less than 1 million weekly users more recently. This seems small next to the hundreds of millions using talking AI.
Claude Code and OpenClaw work differently than browser agents. They use text commands in a terminal. This works better because computers understand text well. Some experts say this method needs 10 to 100 times fewer steps.
Browser agents still improve but face challenges. They need lots of computer power to work. They take pictures of web pages and figure out what to do next. This can be slow and sometimes doesn’t work right.
New companies try different ways to make agents better. One startup uses video instead of pictures to help AI understand. They say this works 50 times more efficiently. They even showed their AI briefly driving a car in San Francisco.