Fujitsu Kozuchi Physical AI is emerging as a significant step toward unifying digital intelligence and real-world operations. Fujitsu has announced the development of Fujitsu Kozuchi Physical AI 1.0, a new technology designed to seamlessly integrate physical AI systems with agentic AI, marking a key milestone in its collaboration with NVIDIA.
At its core, Fujitsu Kozuchi Physical AI aims to move artificial intelligence beyond isolated automation. Instead of AI agents working only within digital boundaries, this platform enables them to collaborate securely, learn continuously, and eventually interact directly with physical environments through robots. The technology integrates NVIDIA’s software stack with Fujitsu’s proprietary AI technologies, reflecting a long-term vision rather than a short-term experiment.
One of the most important features of Fujitsu Kozuchi Physical AI 1.0 is its multi-AI agent framework. This framework allows enterprises to visually design and automate highly confidential business workflows using multiple AI agents.
By leveraging Fujitsu Composite AI, the system automatically combines NVIDIA NIM-compatible services with Fujitsu’s Takane-based AI agents, ensuring both flexibility and maintainability.
Security plays a central role in the platform’s design. Fujitsu has introduced a secure inter-agent gateway that enables AI agents from different vendors to collaborate without exposing sensitive corporate information. This capability addresses a major barrier to enterprise adoption, where data confidentiality and compliance requirements often limit the use of advanced AI systems.
In addition to workflow orchestration, Fujitsu Kozuchi Physical AI introduces specialized AI agents focused on procurement operations. These agents are built on Fujitsu’s Takane large language model and are designed to automate complex purchasing workflows in enterprise environments. The approach reflects Fujitsu’s focus on practical, industry-specific AI rather than generic automation.
The procurement-focused agents perform distinct but connected roles. A document comprehension agent interprets complex procurement documents and converts them into structured data.
A procurement regulation analysis agent examines internal and external rules to generate compliance prompts. Finally, a compliance checking agent verifies adherence to regulations and securely sends requests for quotations to external suppliers.
A proof-of-concept deployment within Fujitsu’s own purchasing department demonstrated tangible benefits. The company reported a reduction of approximately 50 percent in order confirmation workload.
At the same time, compatibility with NVIDIA NIM microservices is expected to boost inference speed by up to 50 percent, significantly accelerating hundreds of daily compliance checks.
Beyond immediate efficiency gains, Fujitsu Kozuchi Physical AI addresses deeper structural challenges facing agentic AI adoption. While AI agents have advanced rapidly, their use in complex, cross-departmental workflows remains limited. Enterprises require AI systems that are not only intelligent but also maintainable, auditable, and secure across organizational boundaries.
The same challenges apply to physical AI, where agents must interact with real-world environments through robots. Such systems demand robust situational awareness, secure data exchange, and an understanding of on-site operations. Fujitsu Kozuchi Physical AI 1.0 is designed to unify these requirements within a single architectural foundation.
Looking ahead, Fujitsu plans to evolve this technology into a full agentic AI foundation by the end of its fiscal year 2025. In this next phase, AI agents will be able to learn and evolve autonomously within customer environments. The platform will then expand further into the physical AI domain, enabling direct collaboration between AI agents and robots in real-world settings.
Fujitsu’s long-term vision is a society where AI agents and physical robots work together seamlessly. In such environments, AI systems would not only analyze data but also execute complex tasks based on a deep understanding of physical operations. This convergence of digital intelligence and physical action represents a major shift in how enterprises may deploy AI in the future.
By combining secure agent collaboration, specialized industry agents, and NVIDIA-powered infrastructure, Fujitsu Kozuchi Physical AI positions itself as a practical foundation for next-generation enterprise automation. As organizations look beyond experimentation toward scalable AI adoption, platforms that unify intelligence, security, and physical interaction are likely to play a critical role.
To stay updated on breakthroughs like Fujitsu Kozuchi Physical AI and other developments shaping the AI landscape, visit ainewstoday.org for more in-depth AI news and insights.
