Fujitsu Launches AI Platform to Automate Software Development
Fujitsu has launched a new AI-driven platform designed to automate the entire software development lifecycle. The system is being rolled out across 67 software packages for the company's medical and government customers. The launch represents a significant move toward using agentic AI to manage complex, end-to-end enterprise workflows.
- In a proof-of-concept for the 2024 medical fee revisions, the platform completed a change request in four hours that would have taken three person-months with conventional methods, a 100-fold increase in productivity. - The system is powered by multiple collaborating AI agents using Fujitsu's "Takane" large language model, which was co-developed with Cohere, and agentic AI technology from Fujitsu Research. - This platform is an evolution of Fujitsu's "Kozuchi" AI services, which include capabilities for generative AI, automated machine learning (AutoML), and AI trust to verify the fairness of AI judgments. - A key component of the strategy is "AI-Ready Engineering," a process focused on preparing a company's existing systems, data, and knowledge so the AI can accurately interpret legacy environments for reliable automation. - Agentic AI systems differ from passive assistants by being able to reason, plan, and execute multi-step tasks with limited supervision, using tools and context to work towards a goal until it is achieved. - Fujitsu is also developing "Physical AI" to connect these autonomous digital agents with physical robotics, aiming to have AI agents directly manage robot fleets in factories and warehouses. - The company plans to expand the platform beyond government and healthcare into finance, manufacturing, retail, and public services by the end of fiscal year 2026. - To address security and data sovereignty concerns, particularly for government and enterprise clients, Fujitsu offers a dedicated, on-premise version of the platform to ensure confidential data is not exposed externally.
