Fujitsu's Technology Solves Challenges in AI Business Adoption Pioneering the Future of Business with Generative AI and AI Agents
Technology News | 2025-11-27
5 minute read
Generative AI is quickly moving beyond being just a tool for creating text and images to becoming an agent that can carry out tasks autonomously. This shift is raising expectations for how businesses can put it to work. But as adoption grows, concerns that have been present since the early days of the technology, like hallucinations or high computing resource demands, are becoming major challenges. For AI to deliver real, sustainable value in the enterprise, these issues now need to be addressed.
This article introduces Fujitsu's latest technologies that solve these challenges, for enterprises seeking to deeply integrate generative AI and AI agents into their business operations.
Expanding Use Cases Through Ultra-Lightweight Generative AI
As the generative AI market continues to grow, the demand for greater computing power has become a serious challenge. Larger models may deliver higher performance in general, but they often fall short when it comes to meeting the specific needs of individual business applications. They can also be difficult and costly to customize.
To overcome these issues, Fujitsu is focusing on developing smaller, domain-specific generative AI models designed to make it easier for businesses to adopt and tailor to their needs.
Fujitsu’s breakthrough “1.0-bit Quantization Technology”, introduced in September 2025, compresses information in Large Language Models (LLMs) to make them lighter and more efficient. Using quantization (the term is unrelated to quantum computing) Fujitsu has achieved size reductions of large language models (LLMs) of up to 94% while still maintaining their accuracy. It has applied the technology to “Takane”, it’s own LLM. Fujitsu is the first company in the world to successfully compress 32-bit and 16-bit data down to just 1 bit. This dramatic reduction in model size delivers an equivalent decrease power consumption, opening up new possibilities for generative AI, from robotics to edge applications.
Fujitsu has also developed “Specialized AI Distillation Technology”, which fine-tunes AI model structures in a way similar to how the human brain strengthens important knowledge and filters out what’s unnecessary. This not only makes models smaller and faster but also improves their accuracy for specific tasks, allowing businesses to customize and optimize AI performance for their unique use cases.
New Insights from Data Structuring × Advanced Analysis
For generative AI to deliver real business value, it must make full use of corporate data. Yet 90% of corporate data is unstructured, making it difficult to extract actionable insights without systematic organization.
Fujitsu’s “Knowledge Graph Enhanced RAG” technology leverages Takane to automatically structure corporate data into knowledge graphs, ensuring generative AI has access to accurate, organized information.
Once data structuring is complete, advanced data analysis becomes possible. Fujitsu possesses extensive intellectual property in the field of "Causal AI," which performs sophisticated analysis of causal relationships between business events to support management decision-making. The combination of data structuring technology and causal AI makes new discoveries and insights possible. For example in retail, identifying the relationship between frequency of visits to stores that customers make and their average spend. Or in medicine, understanding the relationship between genes and lifestyle habits. [1]
Furthermore, “graph AI”, AI trained on graph structures using a large language model, can analyze highly complex genetic networks, uncovering novel treatments for cancer and other diseases.
Building Secure, High-Quality AI Agents Through Domain Specialization and Autonomous Improvement
As generative AI evolves, business applications are emerging that harness AI agents autonomously executing tasks using multiple generative AI models. Fujitsu’s “Multi-AI Agent Framework” specialises AI agents by domain, and enables this complex workflow to be improved in quality and secured.
The key to leveraging AI agents in business lies in the technology enabling them to perform tasks collaboratively with multiple agents within a company's real-world environment. The Multi-AI Agent Framework provides functions to execute workflows while enabling knowledge from humans, robots, and other agents to grow within specific domains. For example, functions which optimize collaboration between agents by enabling negotiation, and functions that securely execute workflows by monitoring agent interactions.
Building on the Multi-AI Agent Framework, Fujitsu has developed the Enterprise AI Agent Platform, a trial-ready foundation for its advanced agent technology, and is now offering a trial environment through Fujitsu Kozuchi.
Strengthening Joint Research in the Sovereign AI Domain
Finally, to help businesses achieve real benefits through generative AI, Fujitsu is accelerating the deployment of its latest developed technologies into customer environments. Specifically, Fujitsu plans to advance joint research and proof-of-concept initiatives with governments and companies in the sovereign AI domain, where AI systems are fully contained within domestic or proprietary infrastructure. The insights and outcomes from these efforts will be integrated into Fujitsu Kozuchi, allowing many other organizations to use and benefit.
For those who want to learn more about AI agent concepts, use cases, and foundational technologies
For those who want to learn more about core generative AI technologies and their business applications
- Domain Specific AI powered by Takane [FRP]
- Self-improving Amalgamation AI [FRP]
- Data-Driven Decision Making [FRP]
- LLM Vulnerability Scanner and Guardrails[FRP]
- HPC & AI fusion technology for zero-emission materials development [FRP]
- Introducing "Fujitsu Causal AI" (Total of 3 Parts) #1 Causal Action Optimization Technology [TECH BLOG]
Working with agentic AI and its potential risks
AI Agents and the Pathway to Evolving Intelligent Manufacturing
Delivering Value in a World with AI Agents: A Fujitsu and Microsoft Collaboration
