Google I/O floods products with Gemini

Google is no longer pitching AI as a feature bolt-on; it is trying to make Gemini the operating layer for its consumer and developer ecosystem. That is a strong strategic move, but it also raises the bar on reliability, cost, and how much autonomy users will actually trust.

• –Gemini 3.5 Flash and Gemini Omni split the stack between fast agentic execution and richer multimodal generation, which is the right shape for real product usage
• –Antigravity, Managed Agents, and AI Studio updates show Google is pushing from prompt-based tooling toward builder infrastructure for agents
• –Search, Workspace, Shopping, and Android integrations give Google a distribution advantage that pure model vendors do not have
• –The new $100 Ultra tier suggests Google is planning for heavier power-user and agent workloads, not just casual chat
• –This feels like a direct answer to the broader agent-platform race, with Google trying to own both the models and the surfaces where they act