Korean Physical AI startup building foundation models for robotic hand manipulation — creator of the RLDX-1 model built on the Multi-Stream Action Transformer (MSAT) architecture.
RLWRLD Inc. is a Korean Physical AI startup founded to build foundation models for robotic hand dexterity (dexterous manipulation). The company operates under the slogan "Dexterity is intelligence" and positions its technology as a universal AI brain for diverse robotic platforms, including K-robot humanoids (e.g., ALLEX). Its key product is RLDX-1 — a foundation model presented as the first "dexterity-first" solution for five-finger robot hands, built on the proprietary Multi-Stream Action Transformer (MSAT) architecture. RLWRLD also runs the open dexterous-manipulation benchmark DexBench. Its headquarters is in Seoul (Gangnam-gu), with operational offices in San Francisco and Tokyo. In April 2026 the company announced a USD 26 million funding round.
Founders
Co-founder and CEO of RLWRLD; known in Korea as 류중희 (Ryu Jung-hee).
Classification
