Google rolls out Project Mariner, its web-browsing AI agent

Google announces time Google I/O 2025 It is rolling out the project mariner, the company’s experimental AI agent that browse and uses websites to more users and developers. Google also says that the project mariner has been updated significantly, allowing the agent to take about a dozen jobs at once.

Our new customers of Google 9 249.99-AI Ultra Plan in the month Get access to the project mariner, and the company has said that support for more countries is coming soon. Google also says that it is bringing the project marinade capabilities to the Gemi API and Vertex AI, allowing developers to create applications powered by agents.

Exposed to the end of the first 2024, Project mariner Users still present Google’s bold efforts to rebuild how users communicate with the Internet through AI agents. During the launch, Google search leaders said that they saw the project mariner as part of the shift of a basic user experience, so that people would instead of instead of visiting websites and completing those tasks to an AI agent more work.

For example, users of the project mariner can buy a baseball game ticket or buy online grocery websites online-they only chat with Google’s AI agent and go to the websites and take action for them.

The project competes with other web-browsing AI agents of the project Marina Operator, Amazon’s Nova ActAnd Ethnographic computerThe These tools are all in an experimental stage and TechCrunch experience prototypes have proved to be slow and risk of mistakes.

However, Google says it has taken a response from the primary examiners to improve the project mariner’s ability. A Google spokesman told TechCrunch that the company has updated the project marina to run on the cloud virtual machines, much like OpenAI and Amazon agents. This means that users can work on other projects when the project marina completes the work in the background – Google says the new project can handle up to 10 jobs at Marina.

This update made the project mariner significantly more effective than its predecessor, which was going to the user’s browser. I have mentioned me as Initial reviewThe initial design of the project mariner means that users could not use other tabs or applications on their desktop while working on AI agent. This kind of AI has defeated the intention of an agent – it will work for you, but you couldn’t do anything while working.

In the upcoming months, Google has said that users will be able to access the project marinner in AI mode experience in AI-driven Google search experience. When it is launched, the feature will be limited to Google’s Opt-in Testing Ground Search Labs for search features. Google says it is working to get some of this agent’s flow with ticketmaster, stubhaba, rasi and vagaro.

Separately today, Google has unveiled the initial demo of another agent, known as “Agent Mode”. The company says that this feature combines web browsing with other Google apps as well as other Google apps. Google says that overcomes customers will soon get access to agent mode on desktop.

In this year’s I/O, Google finally seems to be willing to send the agents that have been talking about the years. Project Mariner, Agent Mode and AI Mode all seem ready to change how users navigate the web and how vendors contact their customers online. Web-browsing agents have a big impact on the Internet economy, but Google seems to be ready to keep these agents out of the world.