Google used its annual I/O developer conference this year to plant a flag firmly in what its CEO Sundar Pichai called “the agentic Gemini era” — a vision in which AI no longer simply assists users in writing or searching, but actively takes action on their behalf across nearly every Google surface.
Headlining the keynote in Mountain View last night was the debut of two new models. First, Gemini Omni, positioned as a major leap in multimodality, world understanding and editing, capable of generating “anything from any input” — beginning with video — and accepting natural conversational instructions for refinement. Alongside it, Google unveiled Gemini 3.5, with the Flash variant leading the rollout as the first in a new family designed to pair frontier intelligence with the ability to execute complex, multi-step agentic workflows.
The agentic theme ran through nearly every announcement. Google detailed major upgrades to Antigravity, its agent-first development platform, which the company says now allows “anyone to be a builder” by orchestrating agents that act rather than merely suggest. Developers also got an enhanced Gemini API and, notably, native Android “vibe coding” support inside Google AI Studio, which is gaining a dedicated mobile app and tighter Google Workspace integrations.
Consumers were not left out. The Gemini app is becoming considerably more proactive with new round-the-clock capabilities including Gemini Spark and a personalised Daily Brief. In Search, Google announced a new generation of Information agents and shared one-year metrics on AI Mode in the United States, reporting a marked shift from keyword queries to natural-language conversations. A separate update covered the rollout of agentic capabilities across the wider product family, from a refreshed Ask YouTube to deeper Gemini integration in Chrome, which received fifteen distinct updates aimed at powering what Google is now calling “the agentic web.”
On Android, the tech giant introduced Android Halo, which surfaces real-time agent activity directly in the status bar, and previewed Google Pics, a new AI design tool that joins fresh voice capabilities arriving in Gmail, Docs and Keep, alongside updates to AI Inbox.
Creators got their share of news through Google Flow and Google Flow Music, both of which gain new agents, mobile apps and Gemini Omni integration. Pomelli added agents that can draft a brand book and spin up a full website, while Stitch introduced real-time, reflowable layout design. To help users navigate the rising tide of synthetic media, Laurie Richardson announced an expansion of SynthID watermarking and broader adoption of C2PA Content Credentials across Google products.
For paying users, Google One received a substantial refresh, headlined by a new AI Ultra plan starting at S$139.99 per month, with additional benefits flowing to AI Plus and Pro tiers. Google Cloud customers, meanwhile, were promised a smoother path “from initial idea to meaningful impact,” with new tooling aimed at enterprise agent deployment.
In total, Google says it announced 100 things at I/O 2026 — but the through-line is unmistakable. After a decade of AI as a feature, Google is now betting the company on AI as an actor for you. Whether users are ready to hand over the wheel fully is the question the next twelve months will answer.