Google announced during Google I/O 2025 Project Mariner, the company’s experimental artificial intelligence agent is launched that browses and uses websites, and more users and developers. Google also says it has been greatly updated how Project Mariner works, allowing the agent to take nearly ten tasks at one time.
Subscribers 249.99 dollars per month AI Ultra plan You will get access to Project Mariner, and the company says support for more countries will come soon. Google also says it brings the Project Mariner capabilities to Gemini API and Vertex AI, allowing developers to build applications supported by the agent.
The niqab was first unveiled in late 2024, Mariner Project Google’s most daring effort is to renew how users interact with the Internet through artificial intelligence agents. At all, Google Research leaders said they saw Project Mariner as part of the transformation of the basic user experience, as people delegate more tasks to the artificial intelligence agent, rather than visiting web sites and completing these tasks themselves.
For example, the Project Mariner users can buy tickets to the baseball game or buy grocery online without visiting a third-party website-they are only chatting with a Google AI agent, visiting websites and taking action for them.

The Mariner project competes with other artificial intelligence agents for the web grades, such as Openai workerand Amazon’s Nova ActAnd The use of human computer. All of these tools are in an experimental stage, and Techcrunch experience has proven that the initial models are slow and vulnerable to errors.
However, Google says notes have been eliminated from the first laboratories to improve the Project Mariner capabilities. A Google Techcrunch spokesman tells that the company is updating the Project Mariner to run on virtual devices in the cloud, such as agents from Openai and Amazon. This means that users can work on other projects while Project Mariner completes the tasks in the background – Google says the new Mariner project can deal with up to 10 tasks at one time.
This Project Mariner update makes it more useful compared to its predecessor, which was working on the user browser. As I indicated in my country Initial reviewThe early design of Project Mariner means that users cannot use tabs or other applications on the desktop while the artificial intelligence agent works. He defeated this type of purpose from the artificial intelligence agent – will work for you, but you could not do anything else while working.
In the coming months, Google says users will be able to reach Project Mariner in artificial intelligence mode, a Google search experience on behalf of artificial intelligence. Upon launch, the feature will be limited to search laboratories, Google Test Land for Search Features. Google says she works with Ticketingmaster, Stumbhub, Resy and Vagaro to run some of these agent flows.
Separately today, Google revealed an early experimental offer for the experience of another agent called “the agent”. The company says this feature combines web browsing with search and integration features, as well as with other Google applications. Google says that Ultra will be able to reach the agent’s position on the desktop soon.
In I/O for this year, Google finally looks ready to charge the agents you are talking about for years. The Project Mariner, Agent Mode and AI Mode seem to be all ready to change how users move on the web, and how sellers interact with their customers online. Web browsing agents have great effects on the Internet economy, but it seems that Google is ready to put all these agents in the world.
https://techcrunch.com/wp-content/uploads/2018/09/GettyImages-956088114.jpg?w=1024
Source link