RLWRLD is advancing the training of physical AI systems by collecting real-world dexterity data using MANUS gloves. This initiative focuses on capturing detailed human hand and finger movements to enhance AI's ability to replicate complex physical tasks. The data gathered aims to improve the precision and adaptability of AI systems in handling intricate manual operations, potentially revolutionizing fields that require fine motor skills.