Transcending human cognitive limitations represents a critical frontier in LLM training. Proprietary agentic systems like DeepResearch have demonstrated superhuman capabilities on extremely complex information-seeking benchmarks such as BrowseComp, a feat previously unattainable. We posit that their success hinges on a sophisticated reasoning pattern absent in open-source models: the ability to systematically reduce extreme uncertainty when navigating vast information landscapes. Based on this insight, we introduce WebSailor, a complete post-training methodology designed to instill this crucial capability. Our approach involves generating novel, high-uncertainty tasks through structured sampling and information obfuscation, RFT cold start, and an efficient agentic RL training algorithm, Duplicating Sampling Policy Optimization (DUPO). With this integrated pipeline, WebSailor significantly outperforms all open-source agents in complex information-seeking tasks, matching proprietary agents' performance and closing the capability gap.

All of which makes AI's machine learning an ideal candidate for massive parallel processing.

Large-area electronics (LAE) is an emerging technology for electronic device manufacture, such as printing or large-scale lithography, the process used to create flat panel displays and solar cells. Using LAE processes, engineers could create systems that are large (several square meters squared) and highly flexible; for instance, based on paper or plastic. Over the past decade or so, many teams worldwide have been working on LAE systems. This has led to the creation of numerous innovative devices, such as large, flexible and sensing artificial skins for robots. Researchers at Princeton University have recently realized a new wireless system based on LAE technology that can operate at gigahertz frequencies.