Google introduced Gemini 3.5 Flash and the most significant redesign of its Search engine in 25 years [1] on May 19, 2026 [2].
These updates represent a fundamental shift in how the company integrates generative AI into its core products. By transitioning to a world-model approach and deploying more efficient AI agents, Google aims to maintain its dominance in the search market against emerging AI competitors.
The company said that Gemini 3.5 Flash is now the default model across both Google Search and the Gemini app [1]. This move streamlines the user experience by providing a faster, more responsive AI interface for millions of daily queries.
Alongside the model update, Google debuted Omni, a world model capable of generating editable video based on user inputs [1]. While some reports describe Omni more broadly as a world model, other details highlight its specific ability to create and modify video content [1].
Google also launched Spark, a new AI personal assistant designed to handle complex tasks. These announcements were part of a keynote presentation that lasted nearly two hours [3].
The most striking change is the overhaul of Search, which the company said was its biggest redesign in 25 years [1]. The new interface is intended to move beyond simple link-sharing toward a more agentic experience that can synthesize information, and perform actions on behalf of the user.
These developments are part of a broader strategy to position Google at the forefront of the generative AI era [4]. By embedding these tools directly into the ecosystem, Google is attempting to transform the internet's primary gateway into an active AI collaborator [4].
“Search received its biggest redesign in 25 years”
The transition to Gemini 3.5 Flash as the default model suggests Google is prioritizing latency and efficiency to make AI-driven search feel instantaneous. By combining a redesigned search interface with the Omni world model and Spark assistant, Google is shifting from a 'search engine' to an 'action engine,' attempting to capture the user's entire workflow rather than just the initial discovery phase.
