Google deployed an agentic AI system to assist with peer review across ICML and STOC, processing approximately 10,000 papers. A formal research paper documenting the deployment has been published. The deployment signals institutional confidence in AI systems for high-stakes gate-keeping functions. This validates agentic workflows in contexts where decision quality directly affects research visibility and career outcomes—moving beyond supervised tasks into judgment-intensive infrastructure. The scale (10K papers) demonstrates operational feasibility at conference throughput. For operators: AI-assisted review workflows shift from hypothetical to proven. Human reviewers can now expect agentic pre-screening, summary generation, and recommendation support as standard infrastructure, reducing human reviewer load but creating new dependency on AI system calibration. This creates demand for validation frameworks that audit fairness and recall across paper categories. Organizations running similar gatekeeping functions (grant review, hiring, content moderation at scale) now have a reference model for agentic integration. The operational question moves from "should we?" to "how do we audit it?"