Save StorySave this storySave StorySave this story<\/p>\n
Google is shaking up the team behind Project Mariner, its AI agent that can navigate the Chrome browser and complete tasks on a user\u2019s behalf, WIRED has learned. In recent months, some Google Labs staffers who worked on the research prototype have moved on to higher-priority projects, according to two people familiar with the matter.<\/p>\n
A Google spokesperson confirmed the changes, but said the computer use capabilities developed under Project Mariner will be incorporated into the company\u2019s agent strategy moving forward. Google has already folded some of these capabilities into other agent products, including the recently launched Gemini Agent, the spokesperson added.<\/p>\n
The change comes as Google and other AI labs rush to respond to the rise of highly capable agents like OpenClaw. While these tools are mostly used by developers today, Silicon Valley believes they could soon power general-purpose assistants for people and businesses. Nvidia CEO Jensen Huang compared the buzzy tool to a new operating system for agentic computers. \u201cEvery company in the world today needs to have an OpenClaw strategy,\u201d he said at the company\u2019s developer conference earlier this week.<\/p>\n
Google CEO Sundar Pichai highlighted Project Mariner during last year\u2019s I\/O conference. At the time, browser agents seemed like the industry\u2019s next big bet, with OpenAI and Perplexity launching consumer agents that promised to automate online tasks for users. The agents could click, scroll, and fill out forms on a webpage, much like a human. However, the adoption of these products has struggled to meet industry expectations.<\/p>\n
Perplexity\u2019s Comet browser agent reached just 2.8 million weekly active users in December 2025. Meanwhile, OpenAI\u2019s 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 on a weekly basis, browser agent usage essentially amounts to a rounding error.<\/p>\n
Momentum in the AI world has shifted dramatically in the last year toward agents like Claude Code and OpenClaw (whose creator was hired by OpenAI). Unlike web-browsing agents, these systems control computers through the command-line, which has proven to be a more reliable way to complete tasks. Some of these products include computer use as a feature, among other agentic abilities. By comparison, browser agents now seem somewhat limited as a stand-alone product.<\/p>\n
Kian Katanforoosh, CEO of the AI upskilling platform Workera who lectures about AI at Stanford, says part of the reason computer use agents haven\u2019t taken off is because of their massive computational requirements. Most of these agents work by taking a series of screenshots of a webpage, feeding that into an AI model, and then taking actions based on what they see. Processing all of that information can be slow and at times unreliable.<\/p>\n
\u201cWhat Claude Code and OpenClaw showed was that it\u2019s actually much more efficient to work with the terminal, because the terminal is text-based and LLMs are text-based,\u201d Katanforoosh said. \u201cIt\u2019s probably 10 to 100X less steps to get to the same outcomes.\u201d<\/p>\n
This isn\u2019t to say that browser agents aren\u2019t improving, or that research into computer use has hit a dead end.<\/p>\n
Last month, the startup Standard Intelligence released a computer use model trained on videos, rather than screenshots. The startup says it developed a video encoder that can compress videos into an AI model\u2019s context window, which it claims is 50X more efficient than previous computer use models. To show off its AI model\u2019s capabilities, the startup hooked it up to a car, a live videofeed, and a computer keyboard. The model was able to briefly drive autonomously around San Francisco.<\/p>\n
Ang Li, CEO of the computer use agent startup Simular and a former Google DeepMind researcher, argues that computer use agents fill a critical gap in agent capabilities, and will likely always be necessary.<\/p>\n
\u201cI do see there always being an 80\/20 split. You can use the terminal to solve a lot of problems already, but there will always be problems you have to solve in the GUI (graphical user interface),\u201d he said. \u201cFor example, if you want to go to a health care insurance website, or some other legacy software, they often don't have an API that a terminal agent can just call up.\u201d<\/p>\n
Still, AI labs broadly seem to be shifting their bets from computer use agents to coding agents. Even for tasks that don\u2019t involve coding, AI labs have found that a coding agent\u2019s ability to use other applications, modify files, and create bespoke software can make them more helpful to users. For example, if someone needs help creating a budget, they can upload bank statements into a coding agent and have it create a custom dashboard to help the user assess their spending habits.<\/p>\n
OpenAI executives say they\u2019d like Codex to power general-purpose agents inside of ChatGPT. Meanwhile, Anthropic has already built a version of this, Claude Cowork, an offshoot of Claude Code that doesn\u2019t require users to open up a terminal. Perplexity, which bet heavily on browser agents, recently launched a similar product called Personal Computer.<\/p>\n
While coding agents have taken off with developers, it\u2019s unclear whether added capabilities will increase adoption among regular people. Google and OpenAI have said consumers could use AI agents to order groceries from Instacart or book a dinner reservation. While those things certainly sound convenient, people may not want to automate such tasks until they\u2019re sure their agent won\u2019t make a mistake.<\/p>\n","protected":false},"excerpt":{"rendered":"
Save StorySave this storySave StorySave this story Google is shaking up the team behind Project Mariner, its AI agent that can navigate the Chrome browser and complete tasks on a user\u2019s behalf, WIRED has learned. In recent months, some Google Labs staffers who worked on the research prototype have moved on to higher-priority projects, according […]<\/p>\n","protected":false},"author":1,"featured_media":47950,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[36],"tags":[],"class_list":{"0":"post-47949","1":"post","2":"type-post","3":"status-publish","4":"format-standard","5":"has-post-thumbnail","7":"category-business"},"_links":{"self":[{"href":"https:\/\/agooka.com\/news\/wp-json\/wp\/v2\/posts\/47949","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/agooka.com\/news\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/agooka.com\/news\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/agooka.com\/news\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/agooka.com\/news\/wp-json\/wp\/v2\/comments?post=47949"}],"version-history":[{"count":0,"href":"https:\/\/agooka.com\/news\/wp-json\/wp\/v2\/posts\/47949\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/agooka.com\/news\/wp-json\/wp\/v2\/media\/47950"}],"wp:attachment":[{"href":"https:\/\/agooka.com\/news\/wp-json\/wp\/v2\/media?parent=47949"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/agooka.com\/news\/wp-json\/wp\/v2\/categories?post=47949"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/agooka.com\/news\/wp-json\/wp\/v2\/tags?post=47949"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}