AgentAI's Robot Safety Verification Platform 'RoboGate' Selected for Nvidia Inception

KOSDAQ-listed AgentAI is accelerating its efforts to enter the global AI robotics market by leveraging "RoboGate," an AI robot safety verification platform jointly developed with its subsidiary AgentAI Labs.

AgentAI announced on June 8, 2026, that RoboGate, currently under development by AgentAI Labs, has been officially selected for "Inception," Nvidia's global startup acceleration program.

Nvidia Inception is a program that identifies promising companies in next-generation technology sectors such as AI, robotics, and semiconductors, and supports them with technology development and global business expansion. The company explained that this selection both lays the foundation for entering the global AI ecosystem and acknowledges its technological competitiveness.

Recently, the AI industry has been rapidly evolving from a software-centric structure to "physical AI," which interacts with real-world environments, and to a humanoid robot-centric paradigm. As global big tech companies such as Nvidia and Tesla enter the race to develop AI-powered humanoid robots, the importance of technologies that verify safety and reliability in real environments, beyond simple performance competition, is growing.

RoboGate, which is being jointly developed by AgentAI and AgentAI Labs, is a platform that pre-verifies whether physical robots equipped with AI models can collaborate safely with humans.

A company representative stated, "As the era of physical robots becomes a reality, the technology to verify how safely AI can operate in real spaces is emerging as a core competitive advantage," adding, "With the selection for Nvidia Inception, we will grow RoboGate into the global standard safety verification solution for the AI robotics industry."

Currently, AgentAI Labs is using RoboGate to assess the real-world response capabilities of major global AI robot models.

According to the company, when evaluating Nvidia's humanoid robot AI model series "GR00T" across a total of 240 safety scenarios, the overall pass rate remained at 16.2%. In particular, in the "human proximity safety" category, none of the 100 scenarios were satisfied, resulting in a final verdict of "NOT_READY" for deployment.

AgentAI Labs explained, "Even if an AI model demonstrates high performance in virtual environments, unexpected malfunctions or decision errors can occur in physical spaces where humans coexist," adding, "A prior safety verification system is absolutely necessary before deploying physical robots in industrial settings."

Going forward, the company plans to expand evaluations to a wider range of global AI robot models, including Nvidia's GR00T series, and to develop customized safety scenarios tailored to the needs of different industrial sites to further advance its verification technologies.

AgentAI stated, "In order for physical robots to become widespread across industries and daily life, ensuring safety and reliability is paramount," adding, "We aim to become a leading technology company in the era of AI-powered physical robots, with RoboGate as our foundation."

This selection for Nvidia Inception is seen as evidence that AgentAI is expanding its business scope beyond that of a simple AI technology company, moving into the "safety verification" market, which has become a core issue in the era of humanoid robots.

This content was produced with the assistance of AI translation services.