Technology
Google I/O 2026: Gemini Omni Rewrites What AI Can Create — and Search Becomes an Agent
Google's annual developer conference unveiled Gemini Omni, a model that generates video from any input, alongside information agents in Search, a Universal Cart, and the Gemini Spark personal AI. The company's most ambitious I/O in a decade.
By Leo W ·
Google's I/O 2026 developer conference, held this week at the Shoreline Amphitheatre in Mountain View, delivered what the company described as its most significant product announcements in a decade. The keynote centred on a single thesis: that AI has moved from a tool that answers questions to a system that takes action on your behalf. The breadth of announcements — from a new multimodal model family to a redesigned Search experience to a 24/7 personal agent — suggests Google is betting its next decade on making that transition real.
Gemini Omni: Any Input, Any Output
The headline announcement was Gemini Omni, a new model family that accepts any combination of images, audio, video, and text as input and generates high-quality video output grounded in Gemini's real-world knowledge. The initial release is Gemini Omni Flash, which is rolling out to all Google AI Plus, Pro, and Ultra subscribers, as well as through the Gemini app and Google Flow. The model can edit existing videos through natural language conversation — a capability that, until recently, required dedicated professional software.
Alongside Omni, Google released Gemini 3.5 Flash, the first in a new model family combining what the company calls 'frontier intelligence with action.' The model is generally available through Google Antigravity, the Gemini API in Google AI Studio and Android Studio, and the Gemini Enterprise Agent Platform. It is also available in AI Mode in Search and is rolling out globally in the Gemini app. Google said Gemini 3.5 Pro is already in internal use and will be released to the public next month.
Search Becomes an Agent
Perhaps the most consequential announcement for Google's core business was the introduction of information agents in Search. The new system allows users to create, customise, and manage multiple AI agents that operate in the background around the clock, reasoning across the web — including blogs, news sites, and social posts — as well as real-time data on finance, shopping, and sports. Agents send comprehensive updates at the moment they become relevant, with links to explore further on the web. Google described this as 'the era of Search agents.'
Google's information agents operate 24/7 in the background, reasoning across the web and sending updates at exactly the right moment. Search is no longer a query box — it is a standing instruction.
Google also announced Antigravity-powered experiences in Search, which use the agentic coding capabilities of Gemini 3.5 Flash to generate dynamic layouts, interactive visuals, and entire custom experiences on the fly in response to search queries. The company plans to extend this to allow users to build persistent custom tools — dashboards, trackers, and mini applications — directly within Search, starting with Google AI Pro and Ultra subscribers in the United States.
Gemini Spark: The 24/7 Personal Agent
Gemini Spark, announced as a 24/7 personal AI agent in the Gemini app, is integrated with Google's suite of productivity tools including Gmail, Docs, and Slides. As a cloud-based agent, it continues working in the background even when the user's device is closed. Users can set recurring tasks, teach it new skills, and create complete workflows. Google emphasised that Spark is designed to ask for confirmation before performing high-stakes actions such as spending money or sending emails — a design choice that reflects the company's awareness of the trust challenges facing autonomous AI agents.
SynthID and the Watermarking Expansion
Among the less-discussed but significant announcements was the expansion of SynthID, Google's digital watermarking technology for AI-generated content. Since its introduction three years ago, SynthID has been integrated into Google's generative media models and products, watermarking over 100 billion images and videos and 60,000 years of audio assets. Google announced it is now expanding SynthID verification to Search and to Chrome in the coming weeks — a move that could make it significantly easier for users to identify AI-generated content encountered during ordinary web browsing.
The I/O 2026 announcements collectively represent Google's most direct response yet to the competitive pressure from OpenAI and Anthropic. The company's strategy is clear: leverage its distribution advantages — Search, Gmail, Android, YouTube — to make Gemini the AI that users encounter in every context, rather than a standalone product they must seek out. Whether that strategy succeeds will depend on whether the underlying models can match the capabilities of rivals who have fewer legacy systems to integrate.
Topics: Google, Gemini, I/O 2026, AI Search, Gemini Omni