Fujitsu has launched its Kozuchi Physical AI 1.0 in partnership with NVIDIA, addressing a significant hurdle in enterprise AI adoption: the challenge of enabling AI agents to collaborate securely without compromising sensitive information. This innovation comes in response to the pressing need for businesses to streamline their operations while safeguarding critical data.
Enterprises often face a complex scenario: a procurement team receives a vendor contract requiring multiple AI systems to analyze the document—one to read it, another to ensure compliance with government regulations, and a third to verify internal rules. However, connecting these AI systems safely is a daunting task; any misstep could lead to severe repercussions, including lawsuits, regulatory penalties, or damaging headlines regarding data breaches.
Fujitsu’s solution is the “secure inter-agent gateway,” a mechanism akin to a translator at the United Nations that enables different AI systems to communicate effectively without disclosing classified information to one another. This approach allows various AI agents to collaborate on tasks while each remains secure, sharing only what is absolutely necessary, thus preserving the integrity of sensitive data.
As businesses increasingly adopt AI agents and autonomous workflows, the focus has shifted from theoretical applications to practical implementations. Fujitsu’s Kozuchi utilizes specialized AI agents instead of a single broad-spectrum AI, employing its Takane LLM to power three distinct agents for procurement tasks. One agent specializes in deciphering complex legal language, another stays updated on regulatory demands, and the third is dedicated to compliance verification.
In trials within its own purchasing department, Fujitsu reported a remarkable reduction in workload, with order confirmation tasks halved. The integration with NVIDIA’s NIM technology is expected to further enhance processing speed by 50%. Such improvements could enable procurement teams to adhere to deadlines more effectively, ensuring critical purchases do not languish in approval delays.
Unlike many existing AI automation tools that apply security measures after deployment, Kozuchi embeds security into its framework from the outset. This feature is particularly vital for sectors like financial services, healthcare, and government contracting, where data breaches pose existential risks. These industries require efficient AI solutions without sacrificing security, which has often been a trade-off in current technologies.
By the end of fiscal 2025, Fujitsu intends to extend the Kozuchi framework beyond document processing to encompass physical operations, such as controlling robots in manufacturing and warehouse settings. This “Physical AI” will facilitate coordination of tangible tasks while maintaining the confidentiality of proprietary processes.
While achieving a 50% reduction in workload in a controlled setting is commendable, the larger challenge lies in adapting this technology across a diverse range of enterprises, each with unique security protocols, outdated legacy systems, and varying compliance frameworks. Fujitsu now faces the task of demonstrating that its solution can scale effectively and meet the complex needs of thousands of companies.
This initiative stands out among AI announcements by concentrating on a specific, pressing challenge faced by businesses today. As enterprises increasingly seek practical solutions to enhance operational efficiency and security, Fujitsu’s Kozuchi Physical AI 1.0 may provide the necessary tools to navigate the evolving landscape of enterprise AI.
See also
Tether CEO Warns AI Market Dynamics Could Disrupt Bitcoin Prices by 2026
Senate Passes AI Civil Rights Act of 2025: New Regulations Impact GOOGL, MSFT, AMZN, and FB
Tesla’s FSD Launches in UAE by January 2026, Achieves Low Hallucination Rate
Factchequeado Invites Latino Media to Apply for $3,000 AI Training Grants by Jan 12, 2026
Sam Altman: AI’s Next Leap to Superhuman Memory Expected by 2026 Amidst Fierce Competition
