OUR STORY
Ludo Robotics builds the social intelligence that lets everyday people instruct, guide, and collaborate with robots naturally — through speech, gesture, and shared understanding. We want robots that don't just perform useful actions, but understand intent, ask clarifying questions, adapt as situations change, and behave in socially appropriate ways. Ludi 0.1 is our first system: an agentic architecture connecting human intent to robot action.
Our headquarters are in Palo Alto, CA with an additional R&D lab in Seoul, Korea. Our team brings together researchers and engineers across robotics, machine learning, human–robot interaction, and agentic AI. We are especially interested in people who are comfortable working across these boundaries—combining ambitious ideas with the practical urgency of getting systems onto real robots and learning from what happens.
Ludo is backed by KRAFTON, a global video game company best known for the genre-defining worldwide success PUBG: Battlegrounds, with deep experience building full-scale reasoning agents and AI partners in rich digital worlds. Game agents already perceive, reason, act, communicate, and collaborate through an embodied character. We are leveraging that expertise in digital embodied intelligence to bring capable agents into the physical world. The embodiment changes; the core agent loop remains.
This creates a powerful digital-to-physical data flywheel. Game and simulation worlds allow us to generate diverse tasks, interactions, and social situations at scale. Real-world robots ground that intelligence in physical experience. Across both environments, interactions become data that can be reviewed, improved, and used to train the next generation of systems.
Our competitive advantages
Game agents perceive, reason, take initiative, and adapt as events unfold — the same loop a robot runs.
Game AI responds instantly under tight compute and latency limits. So must a robot, without waiting on the cloud.
A convincing AI teammate reads intent, holds context, and responds at the right moment. So must a robot people can guide.
Simulation generates social situations at scale; real robots ground them. Both feed one training loop.