Google DeepMind Unveils Gemini 2.0 And Its Multimodal Projects For The Agentic AI Era

Google, the Big Tech giant, has its hand in many jars. From AI to cloud computing to renewable energy and much more.

Just this week, Google introduced Willow, its latest quantum chip, performed a standard benchmark computation in under five minutes. The same would have taken today’s fastest supercomputers 10 septillion years—or 10,000,000,000,000,000,000,000,000 years!

Just days later, Google announced a major leap forward in artificial intelligence with the launch of Gemini 2.0. The move is accompanied by various Gemini-powered launches, such as Project Mariner, Astra and Jules, along with enhancements to Gemini 1.5 Pro’s Deep Research.

These AI advancements signal Google’s vision to take a lead in the upcoming "agentic era," where AI won’t merely process information but actively execute tasks.

Sundar Pichai, the CEO of Google and Alphabet, shared his enthusiasm about the launches, stating, “If Gemini 1.0 was about organizing and understanding information, Gemini 2.0 is about making it much more useful. I can’t wait to see what this next era brings.”

So, what exactly can we expect from Gemini 2.0 and its standout features? How will they impact the future of AI agents? Let’s explore!
 

What Did Google Reveal About Gemini 2.0?

Google calls Gemini 2.0 its most powerful AI model to date and as the first of its upcoming generation of models designed for this new agentic era. A blog post by Sundar Pichai, the CEO of Google and Alphabet, Demis Hassabis, the CEO of Google DeepMind and Koray Kavukcuoglu, the
CTO of Google DeepMind highlighted how Gemini 2.0 will help create new AI agents that bring the world one step closer to a universal AI assistant.

The Gemini 2.0 AI model is capable of:
 

• Multimodal Processing: Gemini 2.0 can interpret and generate text, text-to-speech, images and audio, enabling dynamic, interactive applications across diverse industries.

• Enhanced Reasoning: The model demonstrates advanced reasoning capabilities, excelling in multi-step problem-solving and delivering highly accurate outputs for complex queries.

• Native Tool Use: Gemini 2.0 can execute code, interact with tools like Google Search and integrate third-party APIs, significantly broadening its business potential.

• Speed and Efficiency: Gemini 2.0 Flash performs twice as fast as Gemini 1.5 Pro, enabling real-time responses for users.

Despite its goal to make models accessible, users can only use experimental versions of Gemini 2.0 and Gemini 2.0 Flash via the Gemini API in Google AI Studio and Vertex AI. Gemini 2.0 will be made generally available in January, along with more model sizes.
 

What Is Project Mariner?

Google also mentioned Project Mariner, an early research prototype of an AI agent built with Gemini 2.0. So, how can it help you accomplish complex tasks?
Mariner, a browser-integrated AI agent, aims to redefine web navigation. It understands text, code, images and forms on a screen and can autonomously browse and interact with websites on users' behalf. Mariner can also perform tasks such as filling out forms, retrieving data and analyzing content.

Project Mariner can also type, scroll or click inside an active tab on your browser. The AI agent also asks users for confirmation before sensitive actions, such as purchasing an item online. With its ability to handle text, code and visual elements directly in the browser via an experimental Chrome extension, Mariner can be a powerful assistant to automate how we navigate the web.
 

What Is Project Astra?

While the idea of creating a universal AI agent is not a new one, Astra represents Google’s renewed vision for a universal AI assistant capable of multimodal interactions and context-based personalization. Astra supports multiple languages and can converse even in mixed languages across accents. It integrates seamlessly with Google’s everyday tools like Maps, Lens and Search.

Astra’s memory-based personalization, with up to 10 minutes of in-session memory, ensures it adapts to user preferences over time, making interactions intuitive and efficient. Sundar Pichai also hinted toward Project Astra’s integration within prototype glasses.
 

What Was Said About Gemini 1.5 Pro's Deep Research

While Gemini 2.0 takes center stage, Gemini 1.5 Pro’s Deep Research – the latest agentic feature in Gemini Advanced – continues to deliver exceptional value for in-depth analysis and reporting. So, what makes it such a big deal?

Well, Deep Research employs a network of Gemini 1.5 Pro models working in unison to analyze long-context information. The system generates structured research plans, creating agents to explore extensive datasets, websites and documents.

This allows Deep Research to generate detailed reports, complete with charts, summaries and source citations. Moreover, users can refine their research plans to ensure the final output aligns with their specific requirements. Using Gemini 1.5 Pro’s automation capabilities simplifies data collection and synthesis, saving hours of manual effort.

Dave Citron, Senior Director of Product Management at Gemini, explained its value, saying, “Deep Research is the perfect tool if you’re an entrepreneur launching a small business and want to quickly gather a competitor analysis and recommendations for suitable locations, or if you’re a marketer researching recent AI-powered marketing campaigns to benchmark for 2025 planning.”

Google is exploring how it can apply Gemini 2.0's spatial reasoning capabilities to robotics and collaborating with game developers to build better virtual worlds. With Google betting on agentic AI, will Gemini 2.0 and its projects be a game-changer? It certainly seems so at the moment!
 

Conclusion

Although Google is doubling down on agentic AI, some concerns were raised and addressed by CEO Sundar Pichai.

He said they recognize the responsibility AI development entails and remain committed to ethical AI development by prioritizing transparency, bias mitigation and user safety. With advanced models like Gemini 2.0, the company is working with the Responsibility and Safety Committee (RSC), its internal review group, to ensure all potential risks are identified and mitigated.

On the other hand, with Gemini 1.5 Pro’s Deep Research and groundbreaking capabilities of Gemini 2.0, Google is redefining the role of AI agents in daily life.

Will these AI advancements transform the way we use AI agents? Can Google take the lead in the nascent agentic AI era?

Let us know in the comments below!