RLWRLD builds robotics foundation models that enable robots to perceive, reason, and act with human-like dexterity in real industrial environments. The South Korean company develops what it calls the "core intelligence layer" for the physical world, training AI systems on actual industrial sensor data rather than simulations to help robots handle complex manipulation tasks in manufacturing and logistics.
Founded in 2024 by Jung-Hee Ryu, RLWRLD connects data providers, compute platforms, hardware partners, and enterprise customers around its 4D+ motion-capture data and physical-world foundation models. The company's RLDX model family combines vision-language understanding with proprioception, tactile, and torque sensing to work across robotic arms, dexterous hands, and humanoid robots in dynamic environments.
Our Review
We've been tracking RLWRLD since they emerged from stealth in April 2025, and honestly, this South Korean startup caught our attention for all the right reasons. While everyone else in robotics is still playing around with simulations, these folks are training their AI on real industrial sensor data from actual factory floors.
What makes RLWRLD different isn't just their approach (though that's refreshing). It's their timing and focus. They're not trying to build the next household robot that'll sit in your garage collecting dust.
Real Data, Real Results
The company's RLDX foundation model family is where things get interesting. Their flagship RLDX-1 model, which they open-sourced in May 2026, packs 8.1 billion parameters and combines vision-language understanding with proprioception and tactile sensing. That's a fancy way of saying it can see, think, and feel its way through complex tasks.
We're particularly impressed by their focus on "4D+ multimodal industrial data." Most robotics companies are still figuring out basic object recognition, while RLWRLD is teaching robots to understand the physics of manipulation in real-time industrial environments.
Following the Money Trail
The funding story tells us a lot about investor confidence. After raising $14.8 million in their initial seed round, they quickly followed up with a $26 million Seed 2 in March 2026. That's roughly $41 million in seed funding, which is substantial for a company that's barely two years old.
The investor lineup includes some smart money: FuturePlay, Global Brain, and ANA Holdings. When an airline holding company invests in your robotics startup, you're probably onto something practical.
Why This Matters Now
RLWRLD's timing feels spot-on. Manufacturing is desperately looking for automation solutions that actually work in messy, unpredictable environments. Their B2B focus on factories, warehouses, and logistics centers puts them right where the demand is hottest.
We're also watching their international expansion closely. Appointing Carl Choi as President of RLWRLD USA in March 2026 signals serious intent to crack the American market, while their existing pilots in Korea and Japan give them a solid foundation to build from.
The robotics foundation model space is getting crowded, but RLWRLD's emphasis on real-world industrial data and dexterous manipulation could give them a meaningful edge. We'll be keeping an eye on how their enterprise partnerships develop over the next year.
Features
AI-driven 3D avatar creation
Real-time interactive features
Integration with metaverse platforms
Tools for brands and creators to customize and monetize avatars
Cross-device compatibility
FAQs
Ploy was founded in 2024.

Dexterity is Intelligence.



















