Fujitsu unveils AI platform to automate legacy systems

Fujitsu has launched an AI-driven software development platform, it says, that can automate the full development lifecycle for regulation-driven changes, cutting work that would usually take months down to hours in some cases.

The platform uses multiple AI agents across requirements definition, design, implementation, and integration testing. It has been used in Japan for software updates required by the 2026 medical fee revisions, which adjust public medical fees and how costs are allocated for procedures.

In a proof of concept tied to the 2024 medical fee revisions, Fujitsu reported that a change request that would typically take three person-months using conventional methods was completed in four hours-what it described as a 100-fold productivity gain.

Legacy systems

The launch comes as many organisations face growing pressure to update older systems in response to regulatory and operational change. Healthcare, financial services, and the public sector often rely on large, long-lived applications where updates can require lengthy analysis, manual coding, and extended testing.

Fujitsu positions the platform as a way to deliver those changes faster and with less manual effort. It is designed to interpret complex, evolving enterprise and public-sector systems and then apply the required modifications arising from legal or business changes.

The approach relies on collaboration between multiple AI agents, each responsible for a stage of the development process. Fujitsu says the agents can complete those stages without human intervention, from initial definition through to integration testing.

Takane model

The platform draws on the Takane large language model, developed by Fujitsu jointly with Cohere. Fujitsu Research also contributed agentic AI technology intended for large-scale software development.

Fujitsu also highlighted "AI-Ready Engineering", which it describes as preparing existing assets and knowledge so AI systems can correctly interpret legacy code and documentation, with an emphasis on reliability during automation.

Large regulated environments often include older programming languages, bespoke integrations, and extensive operational workarounds. These factors can complicate change management and increase verification time. Fujitsu says its platform reduces the verification burden during modifications, which has historically consumed significant engineering effort.

Japan rollout

Fujitsu plans to use the platform to update all 67 medical and government business software products provided by Fujitsu Japan by the end of fiscal year 2026. These revisions reflect legal and regulatory changes that require ongoing updates to software and associated workflows.

Work has already begun on modifications related to the 2026 medical fee revisions in Japan. This forms part of a broader push by technology suppliers to apply generative AI to repetitive engineering tasks, including code maintenance and testing.

Australia context

Fujitsu also framed the platform around two issues facing Australia: modernising essential systems and an anticipated shortfall in technology workers. The Future Skills Organisation projects Australia will need 3.5 million finance, technology, and business professionals by 2030, about 450,000 more than today.

Against that backdrop, automated maintenance and testing have become a frequent focus for large employers and service providers. Faster turnarounds on compliance work can also reduce operational risk from delayed updates, particularly in sectors with strict regulatory deadlines.

Mahesh Krishnan, chief technology officer of Oceania at Fujitsu, said customers often see older systems as an obstacle to change.

"Australian business and tech leaders tell us their legacy software is holding them back. Updating the ageing software is risky and expensive. Fujitsu's AI-powered software development platform allows organisations to automate its entire software development process at speed. Importantly, it also addresses our tech skills gap by automating repetitive work, so developers can focus on the high-value innovation that drives Australia forward," said Mahesh Krishnan, chief technology officer of Oceania, Fujitsu.

Commercial plans

Fujitsu is rolling the platform out internally across its software portfolio first, then plans to offer it as a commercial service by the end of 2026.

It also intends to expand the platform's use across finance, manufacturing, retail, and public services by the end of fiscal year 2026.

Fujitsu expects AI-driven development to change engineering work patterns and resourcing. It also plans to strengthen its Forward Deployed Engineer team as it rolls out the approach more broadly.