LivSyn Robotics Raises Series A to Link Robots With AI Models
Beijing-based LivSyn Robotics has raised a Series A round of at least RMB 100 million, the company announced on October 8. The financing brought together a group of financial investors, listed energy-services firm Suwen Electric Energy, and an undisclosed strategic investor focused on embodied AI.
The size and mix of the round underscore continued investor interest in China's embodied AI sector, where startups are racing to build the software layer that connects physical robots with rapidly advancing foundation models. LivSyn did not disclose the precise valuation or the identities of all backers, but the participation of an energy-services company suggests potential synergies with industrial and infrastructure customers.
The company is based in Beijing and works at the intersection of robotics hardware, AI models, and autonomous agents. Its funding announcement comes as Chinese and global investors increasingly look beyond chatbots and software-only AI, betting that the next wave of value creation will involve machines that perceive, reason, and act in the physical world.
According to the October 8 announcement, the round included both financial and strategic investors. Suwen Electric Energy, a listed company in the energy-services sector, is the only named corporate investor so far. The undisclosed strategic investor is described as active in embodied AI, a term that covers robots and other physical systems equipped with AI-driven perception, planning, and control.
LivSyn's core product is RUDA, short for Robotics Unified Device Architecture. The platform is designed to connect different robot designs with AI models and agents, acting as a common layer between heterogeneous hardware and the software that drives it.
The company says RUDA addresses a persistent bottleneck in robotics: most AI models and training pipelines are built for specific machines, so moving a learned skill from one robot to another often requires rebuilding the entire stack. By providing a unified architecture, LivSyn aims to make robot skills and data more portable across form factors and manufacturers.
A key component is PhiAgent, an engine that converts human demonstration videos into robot training data and motion trajectories. Instead of hand-coding every motion, the system can learn from videos of people performing tasks, then translate those demonstrations into instructions that a robot can execute.
Another component, RoboAgent, handles task execution and feeds the results back into the system. That feedback loop is intended to improve performance over time and to help the platform accumulate reusable skills. Together, PhiAgent and RoboAgent form a pipeline that spans data generation, training, deployment, and feedback.
LivSyn says the architecture is meant to let data and learned skills carry over between different robots. If successful, that would reduce the need to rebuild training and deployment systems for each new machine, a costly and time-consuming process that has slowed robotics adoption in many industries.