GroundSubscribers only 57 min ago3Add to bookmarks

Hitachi announces the extension of AI agents to its entire system development process. A strong signal for Japanese integrators and a real-world test for agentics in corporate IT environments.
Hitachi announces (Nikkei Asia, July 24, 2026) the extension of AI agents to all of its system development process - requirements, specification, design, code, testing, and operation. The group targets its enterprise systems divisions (SI, industry, infrastructure), which constitute an important part of the Digital Systems & Services division. This is one of the first public commitments by a historical Japanese integrator to switch from agent-tools to agent-processes, at the scale of its delivery.
Nikkei reports the announcement without publishing unit gain measures per phase at this stage. The promise highlighted: shortened deadlines and reduced workload for engineering teams, with agentic orchestration that coordinates the previously siloed phases. Hitachi does not publicly detail the model stack (frontier partners, latency, internal Lumada platform); the breakdown by type of client project is also not specified.
Hitachi is part of the historical Japanese SIs alongside Fujitsu, NEC, and NTT Data. Unlike the American SI market (Accenture, Deloitte) which has been talking about an agentic shift since 2024, the Japanese majors have been advancing cautiously. The announcement marks a change in language. It's also an implicit admission: the tariff pressure from Japanese clients and the demographics of engineering (mass retirements, difficulties in renewal) make the status quo untenable. To be linked with the Fujitsu / Yaskawa approach (file japan-ai-industrial-stack) which targets physical AI - Hitachi chooses the software and operational path.
Strong signal for the integrated SI segment in Japan: the margin to defend shifts from "billed hours" to "delivery at SLA". For an investor, Hitachi becomes a case study to evaluate whether traditional integrators can absorb AI pressure without collapsing their cost base. For a CIO in Asia, the question becomes: which bricks of my SI foundation can switch to this model without exposing my critical projects?
Hitachi does not publicly detail the model layer, security safeguards, or agent governance in the client environment. A failed visible pilot - a fortiori on a Japanese government project - would durably weigh on the group's trajectory and on the image of the Japanese SI sector.
The communication of Fujitsu, NEC, and NTT Data by the end of Q3 2026: they will be the ones who will be obliged to follow publicly, or to stay in the background - a choice that will redistribute the Japanese SI map.
Create a free account to access all our content and the weekly review.
Article produced by artificial intelligence, reviewed under human editorial control.
Sign in to join the discussion.
I wonder how Hitachi will ensure the reliability and accuracy of AI-generated code and designs.
I'm interested in how Hitachi will manage the integration of AI agents with their existing human workforce. Will there be a shift in roles or collaboration dynamics?
I'm curious how Hitachi plans to handle the ethical implications of using AI agents throughout their development process.
La fin de l'ère SaaS ? Agentic + dette technique