Hitachi and Google Cloud advance OT AI solutions

Under a strategic alliance Google Cloud and Hitachi Ltd, have been working to improve productivity through artificial intelligence (AI) adoption, software modernisation, and enhanced customer service. Hitachi is now extending these efforts into the operational technology (OT) domain, with a focus on developing and deploying AI agents to support frontline workers in social infrastructure sectors such as energy, rail, and manufacturing.

Gemini, Google Cloud’s multimodal AI platform, can process multiple data types simultaneously,including text, images, audio, and video. This makes it particularly suited to frontline environments where tasks often require accurate interpretation of physical information, such as equipment status or product condition. Through Gemini Enterprise, users without specialist expertise can create AI agents using no-code interfaces, thereby streamlining workflows.

To facilitate this, Hitachi has established the "Agent Factory", an AI agent development environment. This environment is being enhanced using Google Cloud technologies to allow for faster and more accessible development of AI agents tailored to specific operational tasks.

By combining Google Cloud’s AI capabilities with Hitachi’s domain knowledge and data governance expertise, the development environment supports the secure handling of complex and sensitive operational data. This is expected to improve the adoption of AI in frontline roles, where implementation has historically been limited. The application of Gemini in such settings marks a significant development.

Internal deployment and future strategy

Hitachi intends to pursue a "Customer Zero" strategy by initially applying AI agents within its own operations. The goal is to build internal use cases across the Hitachi Group using Google Cloud services, and to promote the broader adoption of AI agents through no-code tools that empower employees to develop solutions independently.

GlobalLogic Inc, a subsidiary of Hitachi and a strategic partner of Google Cloud, will play a key role in this process. As a launch partner for Gemini Enterprise, GlobalLogic is actively supporting its deployment both within the Hitachi Group and externally.

Hitachi aims to accelerate value creation by integrating GlobalLogic’s technological capabilities and practical expertise. Through the wider democratisation of AI agents, the company seeks to enable innovation from within operational teams, eventually offering proven solutions to customers facing similar challenges. These initiatives are expected to contribute to more sustainable and safer social infrastructure.

Context and industry challenges

Field-based operations such as maintenance and manufacturing are facing mounting pressure due to labour shortages and an ageing workforce. Ensuring the transfer of technical skills has become increasingly urgent. Hitachi’s strategic alliance with Google Cloud, first announced in May 2024, aims to address these issues by combining Hitachi’s domain expertise with Google Cloud’s AI tools and technologies to drive innovation and productivity.

Jun Abe, Executive Vice President of Hitachi, Ltd. and General Manager of the Digital Systems & Services Division, commented, “Hitachi regards AI not simply as a tool, but as a partner that enhances human capabilities. The progress of our strategic collaboration with Google Cloud has contributed to improved productivity and accelerated innovation. We are now working to extend these outcomes into the operational technology domain, enabling frontline workers to collaborate with AI and focus on higher-value activities. Under our ‘Inspire 2027’ vision, we view the evolution of Lumada to version 3.0, through the integration of AI and domain knowledge, as a key element in building a more harmonised society.”

Hiroyuki Koike, Managing Director of Customer Engineering at Google Cloud Japan, added, “Addressing complex business challenges with generative AI requires not only advanced technology but also the expertise to implement it effectively. Through our strategic partnership with Hitachi, we aim to provide organisations with the necessary tools and guidance to support each stage of their AI adoption.”

Use case: Inspection support at Hitachi power solutions

As part of its initial implementation, Hitachi has launched a proof-of-concept project at Hitachi Power Solutions, a group company specialising in maintenance services for the energy and industrial sectors. The project focuses on inspections of vacuum circuit breakers (VCBs), key components within power distribution panels. These inspections are traditionally labour-intensive and susceptible to human error, such as incorrect bolt tightening, misconnected wiring, or failure to remove discharge clips.

To address these issues, Hitachi is testing AI agents powered by Gemini at its maintenance training centre. The AI agents support restoration checks that were previously conducted using paper-based checklists. Maintenance staff upload before-and-after images of components, and the AI agent analyses the images to detect anomalies and alert users to any issues.

This technology is being evaluated for potential deployment across thousands of annual inspections. It is expected to improve double-checking accuracy, reduce oversight, and enhance both operational quality and efficiency.

The AI agent was developed using a combination of natural language prompts (describing key inspection points and known failure modes) and a dataset of annotated images. While initial configuration may require support from AI specialists, the system is designed to be accessible for frontline staff to make adjustments, thereby lowering barriers to adoption.

Looking ahead, Hitachi plans to explore AI agents capable of analysing video data in real time to monitor adherence to work procedures, offering immediate alerts when deviations or safety risks are identified. The company is also examining the potential for automated report generation from video footage, which could aid knowledge transfer and further reduce manual workload.