In a direct benchmark against human reviewers, the agentic AI tool Loon Lens outperformed Catchii during systematic review title-and-abstract screening. The reported evaluation frames the tool’s performance as exceeding that of trained human screening on the specific task of deciding which records to advance for full-text review. For evidence-generation workflows, systematic screening remains a time-intensive bottleneck. A system that can exceed human performance may reduce manual workload while increasing throughput for rapid reviews. Key follow-ons include robustness across domains, explainability for auditability, and governance around model-driven decisions to avoid missed records or over-inclusion.