(Google)

<이미지를 클릭하시면 크게 보실 수 있습니다>

Google LLC unveiled a demo video of its artificial intelligence (AI) agent Project Mariner on Wednesday local time, which is based on Gemini 2.0.

Project Mariner specializes in searches and information gathering, automating research tasks traditionally done by employees.

In the demo video, a command was given to “search the websites of the companies listed in the notepad and extract the email addresses of their sales representatives.” As soon as the command was given, an AI agent began searching for the first company listed, visiting its website via Google then located the company’s sales information and extracted the email address. Once it completed this task, it moved on to the next company and compiled a contact list of sales representatives’ email addresses in an instant.

Google also showcased Project Astra, an AI agent capable of real-time conversations via smart glasses.

AI agents developed by big tech companies are advancing rapidly. Beyond handling text, they now feature multimodal capabilities, enabling them to understand and respond to visual, auditory, and video inputs. They also use core services like search and mapping as if they were human, and the competition to dominate the market is fierce.

Before Google, Microsoft and Amazon.com had already introduced AI agents for internal corporate use. OpenAI is set to launch its own agent, Operator.

Meanwhile, AI agents from South Korean companies remain at the chatbot level and are just beginning to progress from text to handling images, falling far behind multimodal capabilities. The concept of agents thinking and acting independently and taking action remains a distant goal, according to industry insiders.

Critics warn that Korea, which once led in core technologies like search and messaging, risks becoming technologically dependent in the AI era. With big tech companies adapting their AI models to account for national characteristics such as language and culture, concerns are growing that Korea’s AI sovereignty strategies could lose their competitiveness.

“AI agents require not only advanced technology but also immense funding, making it a challenging field for Korean companies to compete in,” Korea University Professor Choi Byoung-ho said.