Google Unveils Agentic AI Capabilities Across Products at I/O 2025

Google Unveils Agentic AI at I/O 2025

At the Google I/O 2025 developers conference, CEO Sundar Pichai introduced a suite of agentic AI capabilities across Google's product lineup, marking a significant leap in artificial intelligence technology. These advancements aim to provide users with more autonomous and efficient digital assistance 1

1

.

Agent Mode in Gemini App

The centerpiece of Google's agentic AI rollout is the new Agent Mode in the Gemini app. This feature combines advanced AI models with access to tools, allowing the system to perform complex tasks on behalf of users 1

1

. For instance, when apartment hunting, Agent Mode can:

• Search listings on websites like Zillow
• Apply specific filters based on user criteria
• Schedule property tours autonomously

This functionality is made possible through the integration of the Model Context Protocol (MCP), which enables seamless interaction between AI tools and web services 2

2

.

Project Mariner and Multitasking Capabilities

Google's Project Mariner, an early research prototype, has evolved to include enhanced multitasking abilities and a "teach and repeat" function. This allows the AI to learn from user demonstrations and apply that knowledge to similar tasks in the future 1

1

. The system can now oversee up to 10 simultaneous tasks, significantly improving its efficiency and versatility 3

3

.

Agentic AI Across Google Products

Google is extending agentic capabilities beyond the Gemini app:

1. Chrome and Search: Implementing AI-driven features to enhance web browsing and information retrieval 2

   2

   .
2. Creative Suite: Introducing Veo 3 for video generation, Imagen 4 for image creation, and Flow for agentic filmmaking assistance 3

   3

   .
3. Gemini 2.5 Pro: Powering advanced reasoning and responsiveness in Google Search and AI Overviews 3

   3

   .

Ecosystem Development and Integration

To foster an agent ecosystem, Google has introduced:

• The open Agent2Agent Protocol for inter-agent communication
• Compatibility with the Model Context Protocol (MCP) in the Gemini API and SDK
• Plans to make Project Mariner's capabilities available to developers via the Gemini API 1

  1

Availability and Pricing

Agent Mode will be available as an early preview for subscribers of the new Google AI Ultra plan, priced at $250 per month in the US. Wider availability details are yet to be announced 2

2

.

Implications and Future Prospects

The introduction of these agentic AI capabilities represents a significant step towards more intuitive and powerful AI assistants. As Pichai stated, "This is an exciting moment for AI. We're just beginning to see what's possible when AI understands more of the world around you -- and can take meaningful action" 3

3

.

These advancements promise to revolutionize how users interact with technology, potentially increasing productivity and simplifying complex digital tasks. However, they also raise questions about data privacy, AI autonomy, and the changing landscape of human-computer interaction.