Google officially announced a technology transition centered on 'agentic AI' by unveiling a large number of next-generation AI strategies at this year's developer event, I/O 2026. AI that goes beyond simple question-and-answer to understand user intent and perform tasks autonomously has begun to be applied across the entire Google ecosystem.

Google held its annual developer conference, 'Google I/O 2026,' at the Shoreline Amphitheater in Mountain View, California, on the 19th (local time).

The most notable change at this event is the complete overhaul of the search bar for the first time in 25 years.

Google introduced the 'Intelligent Search Bar,' designed to refine questions based on various inputs—including not only text but also images, files, and browser tabs—and provide context-appropriate suggestions.

With the addition of a 'search agent' that browses the web 24 hours a day, a feature has also been added that automatically collects information of interest pre-set by the user and provides notifications.

The AI model has also been significantly strengthened.

Google DeepMind has unveiled 'Gemini Omni,' a multimodal model based on a 'world model' that understands the laws of physics.

This model integrates the processing of text, image, audio, and video inputs, and can perform complex tasks such as video editing, background changes, and character insertion using only natural language.

The generated content is automatically marked with a 'SynthID' watermark, allowing for authenticity verification.

Another key announcement is the 'Gemini 3.5 Flash'.

Google emphasized that this model has significantly improved speed and coding performance compared to the previous version, and that its output speed per second is four times faster than competing models.

The upcoming '3.5 Pro' has already passed internal testing. It is known to be in use.

'Gemini Spark,' which serves as a personal assistant, was also unveiled.

Spark automatically handles repetitive tasks in the background, such as organizing emails, analyzing schedules, and summarizing documents, even when the user closes the device.

It integrates with external services to perform tasks such as reservations and orders, but is designed to require user approval for sensitive actions such as payments.

Changes also continued in the hardware sector.

Google unveiled its AI smart glasses, currently under development in collaboration with Samsung Electronics and Gentle Monster.

It will be released in two forms: voice-based 'audio glasses' and 'display glasses' that project information into the field of view, and offers various functions such as real-time translation, navigation, photography, and AI editing.

This product, built on Android XR, is scheduled to be released this fall.

Through this announcement, Google declared a transition beyond its 'AI First' strategy to the era of 'Agent AI,' where AI takes direct action.

As a structure in which AI takes the lead across all domains—from search and models to platforms and hardware—becomes fully established, the view that the central axis of future technological competition will shift to 'active AI' is gaining traction.