Google delivered its most ambitious I/O keynote on 19 May, unveiling over 100 announcements across AI models, developer tools, consumer products, and hardware. CEO Sundar Pichai positioned the event as the moment Google transitioned from an AI assistant company to an AI agent company — systems that do not just answer questions but take action autonomously.
The headline model release is Gemini 3.5 Flash — the first in Google's latest series combining frontier intelligence with action. Flash surpasses the much larger Gemini 3.1 Pro on coding, agentic, and multimodal benchmarks while running four times faster than other frontier models in output tokens per second. It is generally available immediately via Google Antigravity, the Gemini API in Google AI Studio, and Android Studio.
Antigravity 2.0, Google's agent-first developer platform, received its most dramatic demonstration yet: the team built an entire operating system on stage and then ran Doom on it. The new standalone desktop app provides a centralised workspace to steer, customise, and orchestrate agents, with support for spinning up specialised subagents, cross-platform terminal sandboxing, credential masking, and hardened Git policies. Managed Agents in the Gemini API allow developers to spin up an agent with a single API call that reasons, uses tools, and executes code in an isolated Linux environment.
Gemini Omni represents Google's push into AI-generated video. The new model series accepts image, audio, video, and text inputs and outputs editable video grounded in real-world knowledge. It is rolling out to AI Plus, Pro, and Ultra subscribers through the Gemini app and YouTube Shorts.
Gemini Spark transforms the Gemini chatbot from an assistant into a 24/7 personal agent that runs in the cloud. Launching next week for Google AI Ultra subscribers in the US, Spark integrates Gmail, Docs, and other Workspace tools to perform multi-step tasks autonomously — booking travel, drafting documents, managing email workflows — without requiring the user to be actively engaged.
On the hardware front, Google unveiled Android XR audio glasses — intelligent eyewear launching this autumn in partnership with Samsung, Qualcomm, Gentle Monster, and Warby Parker. Notably, the glasses support pairing with both Android phones and iPhones, signalling Google's intent to reach Apple's user base directly.
For context engineers, Gemini 3.5 Flash's performance-per-token economics and Antigravity 2.0's agent orchestration capabilities represent the most significant competitive response to Claude Code and Anthropic's developer ecosystem. The Managed Agents API — single-call agent creation with isolated execution environments — directly mirrors the patterns that context engineers have been building manually. Google is attempting to make agentic development the default rather than the exception.
