Google Project Mariner: AI Agent for Web Browsing

Google Expands Access to Project Mariner AI Agent
During Google I/O 2025, Google announced the wider rollout of Project Mariner, its experimental AI agent capable of browsing and interacting with websites. This expansion extends access to both users and developers.
Significant updates have also been implemented to enhance Project Mariner’s functionality, now enabling the agent to manage approximately twelve tasks concurrently.
Availability and Access
Access to Project Mariner is currently being offered to subscribers of Google’s new AI Ultra plan, priced at $249.99 per month within the U.S. Google has indicated that support for additional countries will be implemented in the near future.
Furthermore, the capabilities of Project Mariner are being integrated into the Gemini API and Vertex AI platforms. This will empower developers to create applications leveraging the agent’s abilities.
A Shift in Internet Interaction
Initially revealed in late 2024, Project Mariner signifies Google’s most ambitious undertaking to date in redefining user interaction with the internet through AI agents.
Google Search leadership previously positioned Project Mariner as a catalyst for a fundamental shift in user experience. The vision involves users increasingly delegating tasks to an AI agent, rather than directly visiting websites to complete them.
Practical Applications
Users of Project Mariner can, for instance, procure tickets for a baseball game or complete online grocery shopping without ever navigating to a third-party website. The process involves simply communicating with Google’s AI agent, which then autonomously visits websites and executes the necessary actions.
Competition and Current Landscape
Project Mariner enters a competitive arena alongside other web-browsing AI agents, including OpenAI’s Operator, Amazon’s Nova Act, and Anthropic’s Computer Use.
It’s important to note that these tools are currently in experimental phases, and initial testing, as demonstrated by TechCrunch, has revealed prototypes that are occasionally slow and susceptible to errors.
Improvements and Infrastructure
Google states that it has incorporated feedback from early testers to refine Project Mariner’s performance. A spokesperson for Google confirmed to TechCrunch that the agent has been updated to operate on virtual machines within the cloud environment.
This architectural change, mirroring approaches taken by OpenAI and Amazon, allows users to continue working on other tasks while Project Mariner processes requests in the background. Google asserts the updated agent can now handle up to ten tasks simultaneously.
Enhanced Usability
This update represents a substantial improvement over the previous iteration of Project Mariner, which operated directly within the user’s browser.
As previously observed, the earlier design restricted users from utilizing other tabs or applications on their desktop while the AI agent was active, diminishing the core benefit of an AI assistant.
Future Integrations
In the coming months, Google plans to integrate Project Mariner into AI Mode, its AI-enhanced Google Search experience. Initially, this feature will be available through Search Labs, Google’s platform for testing new search functionalities.
Google is collaborating with companies like Ticketmaster, StubHub, Resy, and Vagaro to facilitate these agent-driven processes.
Introducing Agent Mode
Alongside Project Mariner, Google also showcased an early demonstration of “Agent Mode,” another agentic experience. This feature integrates web browsing with research capabilities and connections to other Google applications.
Access to Agent Mode will soon be granted to Ultra subscribers on desktop platforms.
A New Era of Web Interaction
Google appears committed to delivering the agentic experiences it has discussed for several years, as evidenced by the launch of Project Mariner, Agent Mode, and AI Mode.
These advancements collectively have the potential to reshape how users navigate the web and how businesses engage with their online customers. The emergence of web-browsing agents carries significant implications for the internet economy, and Google is actively deploying these agents.
Related Posts

ChatGPT Launches App Store for Developers

Pickle Robot Appoints Tesla Veteran as First CFO

Peripheral Labs: Self-Driving Car Sensors Enhance Sports Fan Experience

Luma AI: Generate Videos from Start and End Frames

Alexa+ Adds AI to Ring Doorbells - Amazon's New Feature
