Hitachi DS: Is Agentic AI Transformative or Another Hype?

Although its early foundations stretch back to the 1950s, the way humans interact and collaborate with AI is taking a dramatic leap forward with agentic AI.

Intelligence that can make decisions and perform tasks without human intervention, agentic AI acts autonomously and is characterised by its goal-driven behavior.

Premkumar ‘Prem’ Balasubramanian is the CTO at Hitachi Digital Services, leading the Technology and Solutions unit and setting the division’s overall technology strategy.

“At Hitachi Digital Services, our overall vision for AI is two-fold: applying AI in every service we provide to accelerate delivery and enhance competitive positioning and designing every solution as an intelligent AI-native solution,” he says.

“We aim to be an AI-native integration partner, delivering future-proof, intelligent architectures built around agentic systems. Our R2O2.ai framework is crucial for building trustworthy AI solutions and fostering enterprise adoption, operationalised by HARC for AI to improve reliability and cost control.”

In light of this, Hitachi Digital Services has launched Hitachi Application Reliability Centers (HARC) for AI, which it calls “a powerful new service designed to help enterprises run AI and Gen AI applications with greater reliability, efficiency and governance”.

This builds on the broader HARC platform the company launched in 2022 to support companies with the challenges they face when scaling AI, which include unpredictable costs, performance degradation and limited oversight of complex models like LLMs.

Following the launch of HARC, Prem shares his thoughts about agentic AI with Technology Magazine.

There is a lot of discussion around agentic AI. Is this another hype or do you truly believe agentic AI is transformative?

I absolutely believe agentic AI is transformative, far beyond mere hype. 

Its true impact lies in its unique ability to handle complex, dynamic situations with unparalleled autonomy and adaptability that traditional automation or basic generative AI cannot.

We’re seeing it deliver tangible business value and measurable ROI, such as reducing deal due-diligence time by 60% for a large private equity firm, improving accuracy of automated invoice processing and reducing cost of processing by over 90%.

Agentic AI significantly enhances system reliability and uptime in critical IT/OT operations, leading to substantial cost savings and operational improvements.

At Hitachi Digital Services, our vision is to be an AI-native integration partner delivering intelligent, agentic architectures. 

The launch of our Center for Architecture and AI (CAAI) and HARC.agents specifically aims to embed AI deep into real-world workflows, augmenting human efforts. 

This signals a decisive shift from pilots to enterprise-wide industrialisation and scaled deployment, proving its lasting value and moving beyond conceptualization to execution.

What do you think are the key success factors for scaled deployment and adoption of agentic AI?

For scaled deployment and adoption of agentic AI, demonstrable business value and ROI are paramount. 

Enterprises must prioritise initiatives that show clear efficiency gains, cost reductions and potential new revenue streams.

Secondly, building trustworthy and responsible AI solutions is crucial. This involves robust data governance, security and reliability frameworks like our R2O2.ai and operational reliability tools like HARC for AI.

Third, evolving to intelligent, AI-native architectures that seamlessly integrate with existing IT/OT systems is vital. This ensures adaptability, future-proofing, and effective deployment.

Finally, success hinges on operational excellence for scalability, focusing on specific, high-value use cases where agentic AI’s unique capabilities truly deliver impact and leveraging expert partnerships for guidance.

What is the core purpose and charter of Hitachi Digital Services’ new Center for Architecture and AI?

We recently launched the Center for Architecture and AI (CAAI) as our next significant step in shaping the future of intelligent architectures. 

Its core purpose is to serve as a structured innovation engine, providing the velocity, clarity and scale necessary to lead in the new era of agentic AI, where AI should be the foundation of modern digital services.

The CAAI’s charter is to embed AI deeply and systematically into next-generation architectures. Our mission is to set a new standard for building intelligent, scalable, production-ready applications that deliver measurable business impact. 

At its core are HARC.agents — modular, domain-specific AI agents designed for seamless Human + AI augmentation, enabling us to go from concept to value rapidly. 

This aligns with our vision to be an AI-native integration partner delivering future-proof, intelligent architectures.

What are some of the challenges your customers are facing when adopting agentic AI?

When adopting agentic AI, our customers frequently face several key challenges:

• Complexity and integration: Agentic AI requires a significant evolution of enterprise IT architecture to seamlessly integrate with existing rule-based logic and IT/OT systems. Customers often struggle to bridge this gap without deep technical expertise.
• Data governance and security: Autonomous agents are heavily data-reliant, raising critical concerns about data privacy, security and compliance. Many organisations aren’t yet fully prepared for the volume, velocity and trustworthiness of data required.
• Building trust and reliability: Enterprises need to ensure AI operates reliably and ethically. Without robust frameworks, there’s a risk of unpredictable costs, performance degradation and limited oversight.

We address these by providing edge-to-core expertise for integration, our R2O2.ai Framework for trustworthy and reliable AI and HARC for AI to ensure operational excellence, observability and cost control.

Can you share a couple of customer examples of deploying AI in production? 

There are two key use cases that I would like to highlight.

Hyper Mobility Asset eXpert

Hitachi Rail’s Remote Command & Control Operations platform is integrating subsystems like automatic train control and communication. 

Using Lumada, it captures and analyses real-time data from various sources, ensuring high observability and reliability. 

Key features include predictive maintenance, fault diagnosis and remote commands. 

The system processes over 25GB of data daily, detecting failures in real-time and reducing maintenance costs by more than 30%. 

This solution is a true One Hitachi solution that’s delivered in partnership with Hitachi Rail, Hitachi Digital, Hitachi Digital Services, GlobalLogic and Nvidia. 

Intelligent data processing

A global packaging leader optimised processing of 1.2 million supplier invoices using Hitachi Digital Services’ Intelligent Document Processing (IDP). 

The new Gen AI solution achieved 91% accuracy, reduced processing costs by more than 90% and streamlined exception handling. 

Where do you see agentic AI heading in the next 12 to 24 months?

In the next 12 to24 months, I predict a decisive shift from agentic AI conceptualisation and pilots to enterprise-wide industrialisation and scaled deployment. 

This acceleration will specifically target use cases that deliver demonstrable and quantifiable business value.

While the ‘silver bullet’ hype around agentic AI will subside, its adoption will grow significantly in areas like hyper-efficient back-office automation, intelligent managed services and predictive operational technologies.

This scaling will be driven by the ability of system integrators to embed agentic capabilities seamlessly into existing complex enterprise architectures.

Crucially, the maturation and widespread adoption of trustworthy and responsible AI frameworks like our R2O2.ai will be paramount as customers prioritise solutions that guarantee reliability, observability and compliance. 

Agentic AI will transition from experimental projects to mission-critical business enablers that generate tangible ROI and, increasingly, unlock new revenue streams. The focus will shift towards practical application and execution over speculative investments.