Google’s latest bet sounds less like a product launch and more like a defensive maneuver. At the center sits Gemini, the company’s frontier model that now powers both a new class of AI agents inside Google Search and a revived smart glasses effort. Instead of simple query matching, search results will be shaped by multi-step reasoning, context retention and task execution that runs across Gmail, Maps and Docs.
The smart glasses move looks like an admission that phones are a bottleneck. Lightweight frames paired with on-device Gemini processing and cloud inference promise real-time translation, scene understanding and voice-first assistance, all routed through Google’s existing index and ad infrastructure. That combination turns glasses from a hardware experiment into a distribution node for Gemini, tightening the feedback loop on user behavior and training data.
The strategy is blunt: Google cannot afford to let Anthropic and OpenAI own both the model narrative and the user interface. By wiring agentic workflows directly into search, the company tries to leverage its query stream as a proprietary moat and keep a closed-loop around discovery, recommendation and commerce. Whether users accept agents that rewrite how search feels, or glasses that put Gemini at eye level, will decide how narrow that competitive gap really becomes.
