In an exclusive conversation with CIO&Leader, Premkumar Balasubramanian, CTO at Hitachi Digital Services, shares insights on business strategy and the emerging challenges brought on by the rise of Agentic AI.
CIO&Leader: The Salesforce report suggests a 383% rise in agentic AI adoption by 2027. How is your organization aligning its business and talent strategy to leverage this shift?
Balasubramanian: That’s an excellent question, and it speaks to the significant shift we’re witnessing in the enterprise AI landscape. While I can’t comment on specific external reports, we at Hitachi Digital Services certainly foresee a decisive shift from Agentic AI conceptualization and pilots to enterprise-wide industrialization and scaled deployment in the next 1-2 years. This aligns perfectly with our strategic direction.
Here’s how we are aligning our business and talent strategy to leverage this profound shift:
Business Strategy Alignment:
- Becoming AI-Native: We are making AI the foundation for modern digital services, integrating it into every service we provide and designing every solution as fundamentally AI-native. Our ambition is to be an AI-Native Service provider and AI-Agent builders of the world.
- Structured Innovation: We recently launched our Center for Architecture and AI (CAAI), a global, cross-functional initiative led by my office, designed to deeply embed AI into next-generation architectures and deliver measurable business impact.
- HARC.agents for Practical Value: At the core of CAAI is HARC.agents, our modular system of prebuilt AI agents designed for rapid deployment and seamless human-AI augmentation in key domains like Industrial, Security, and Operations.
- Future-Proof Architectures: We are advancing intelligent architectures that blend rule-based and reasoning-based logic, ensuring seamless integration of Agentic AI into existing IT infrastructure. Our edge-to-core expertise is crucial here, bridging the gap between IT and OT systems.
- Focus on Tangible ROI & Trust: Our strategy is driven by delivering quantifiable business value, efficiency gains, and cost reduction. We prioritize building trustworthy and responsible AI solutions, operationalizing our R2O2.ai framework through services like HARC for AI to ensure reliability, observability, and cost-effectiveness.
- Evolving Business Models: We are exploring new monetization models, moving towards outcome-based solutions, intelligent managed services, and “Agent-as-a-Service” platforms, positioning ourselves as an AI-Native Integration Partner.
Talent Strategy Alignment:
- Interdisciplinary Collaboration: Our CAAI initiative fosters deep collaboration across our Cloud & Data Services, Enterprise Applications, IoT teams, and Hitachi R&D, actively building interdisciplinary AI talent.
- Deepening Specialized Expertise: We are committed to deepening our capabilities in critical areas like OT, data security, and next-gen architectures, essential for the complexities of Agentic AI deployments.
- Operationalizing AI at Scale: Our focus is on cultivating talent capable of scaling Agentic AI from pilots to enterprise-wide industrialization, emphasizing operational excellence and practical application.
- Knowledge Codification: We are codifying knowledge from our practices into reusable AI assets, fostering continuous learning and development within our global teams.
This comprehensive approach ensures we are well-positioned to lead our clients through this transformative era of Agentic AI.
CIO&Leader: Agentic AI is seen not just as a tool but a strategic lever. What are some high-impact business areas where you plan to deploy agentic AI in the next 12–24 months?
Balasubramanian: As I discussed, we see Agentic AI as a fundamental shift, moving beyond mere enhancement to become the foundation for modern digital services. In the next 12-24 months, my prediction is a decisive shift from conceptualization to enterprise-wide industrialization and scaled deployment. Leveraging our HARC.agents framework, we have identified 6 high-impact business areas where we plan to deploy agentic AI both internally (where applicable) and for our customers. These are:
- Industrial AI – Focused on industries where industry-specific use cases matter the most and where AI sits as the convergence of Information Technology and Operations Technology. This includes work we are doing for customers around predictive maintenance, intelligent data processing, enhanced recommendations engine etc.
- Operations AI – Applying agentic-AI to run operations for our customers. The immediate focus is on application and infrastructure operations. Through the leverage of autonomous agents, we are transforming our existing managed services function into an Intelligent Managed services function.
- Engineering AI – Integrating AI into all phases of Software Engineering life cycle to significantly accelerate go to market for our front businesses and customers to maintain competitive edge and bring their solutions to market at breakneck speeds.
- Analytical AI – While there are a lot of use-cases that benefit from the usage of LLMs, with over 95% of data being held within an enterprise, there is still a lot of scope for building and training custom models for better leverage of enterprise data.
- Security AI – Applying Agentic-AI for security both across IT and OT. With edge AI becoming more applicable, we believe we can enable better security at the edge for our OT customers, which has been a problem for a long time now.
- Cloud AI – Leveraging Agentic-AI for accelerating migration and modernization of on-prem workloads on to the cloud. With the current speed of change, it is even more important for organizations to embrace cloud and leverage AI to accelerate the same is one of our critical focus areas.
CIO&Leader: How are you ensuring that the adoption of agentic AI is not just a technology initiative, but a business transformation led by leadership across functions?
Balasubramanian: I ensure the adoption of Agentic AI is a business transformation driven by leadership across functions through several key approaches:
- Strategic Priority from the Top: Our President and CEO, Roger Lvin, has explicitly named scaling AI adoption as a key strategic priority for FY25. As CTO, I am responsible for setting the technology strategy and building thought leadership in AI.
- Structured Cross-Functional Integration: We’ve launched the Center for Architecture and AI (CAAI) as a cross-functional, global working group. It is led by the CTO Office in close collaboration with Cloud & Data Services, Enterprise Applications, IoT, and strategic partners like Hitachi R&D. This ensures AI is deeply and systematically embedded into next-generation architectures, rather than treated as a mere enhancement or bolt-on capability.
- Focus on Tangible Business Value: We emphasize delivering measurable ROI and concrete business outcomes from AI investments, such as significant efficiency gains and cost reductions. This ensures leadership sees the clear business case beyond just technology, driving adoption in areas like hyper-efficient back-office automation and predictive operational technologies.
- Operationalizing Trust and Reliability: Our R2O2.ai framework (Responsible, Reliable, Observable, and Optimal AI) guides responsible AI development and deployment. HARC for AI then operationalizes this framework, bringing AI observability, lifecycle management, and performance tuning into real-world use to ensure systems are reliable, responsible, and cost-effective, fostering greater confidence and accelerating enterprise-wide adoption by mitigating risks.
- Embedding AI into Core Architectures: We focus on building intelligent architectures that seamlessly integrate Agentic AI capabilities into existing ecosystems, bridging IT and Operational Technology (OT) systems. This ensures future-proof solutions where AI is foundational, not just an add-on.
CIO&Leader: Despite ambitious plans, 88% of Indian firms haven’t implemented agentic AI yet. What are the key barriers your company faces in moving from intention to execution?
Balasubramanian: Despite ambitious plans, we recognize that moving from intention to execution with Agentic AI involves several key barriers, which we address through our strategic approach:
- Complexity and Lack of Understanding: A significant challenge is the inherent complexity of Agentic AI and ensuring a clear understanding of its unique value proposition beyond traditional automation. We emphasize that Agentic AI is not a universal “silver bullet” for all problems.
- Integration with Existing Systems: Seamlessly integrating Agentic AI capabilities into existing, often rule-based, enterprise IT architectures is crucial. This requires a significant evolution rather than a complete re-architecture of operational systems.
- Data Governance and Security Concerns: Autonomous agents heavily rely on data, raising critical concerns around data privacy, security, and compliance. Many organizations are not yet fully prepared to handle the volume, velocity, and trustworthiness of data required for robust deployments.
- Building Trust and Ensuring Reliability: As AI adoption scales, paramount concerns include operational stability, performance degradation, and ethical considerations. Trustworthy AI solutions are essential for enterprise-wide adoption.
- Defining Clear ROI and Measuring Success: Customers increasingly demand measurable returns on investment and concrete business outcomes from AI investments.
- Managing Proliferation and Cost Overruns: The potential for uncontrolled proliferation and management of agents can lead to unexpected cost overruns.
CIO&Leader: With predictions of up to 25% workforce redeployment, how are you rethinking job roles, responsibilities, and career paths in the age of AI augmentation?
Balasubramanian: We are actively rethinking job roles, responsibilities, and career paths in the age of AI augmentation, as our strategy embraces the shift towards intelligent, AI-native solutions. Our approach focuses on:
- Human + AI Augmentation: We are fundamentally moving towards “seamless Human + AI augmentation” where AI is not just an enhancement but a core foundation for modern digital services. This means roles are evolving from purely manual or rule-based tasks to those augmented by intelligent agents that lead end-to-end enterprise workflows.
- Redefining Operational Roles: Agentic AI’s capability to automate labor-intensive processes significantly impacts traditional roles. For example, by implementing agentic processes, we’ve helped clients achieve substantial reductions in human operators (e.g., 70% reduction in invoice categorization) while increasing accuracy. This shifts human roles towards oversight, strategic management, and handling exceptions rather than routine execution.
- Building AI-Native Expertise: Our vision in relation to AI is to be an “AI-Native Service provider and Agent builders of the world”. This necessitates our workforce to develop deep expertise in designing and delivering “future-proof, intelligent architectures”, blending rule-based and reason-based logic. Career paths are increasingly centered on AI architecture, data engineering, and AI lifecycle management.
- Prioritizing Trustworthy AI: With the scale-up of AI adoption, ensuring trustworthiness and responsibility is paramount. Our R2O2.ai framework (Responsible, Reliable, Observable, and Optimal AI) guides development and operationalization. This creates new roles focused on AI governance, ethics, reliability monitoring, and performance optimization.
CIO&Leader: What steps are you taking to future-proof your workforce—reskilling, upskilling, or hiring—for a future where agentic AI becomes a co-worker, not a replacement?
Balasubramanian: To future-proof our workforce for a future with Agentic AI as a co-worker, not a replacement, we are taking several steps:
- Focusing on Human + AI Augmentation: Our strategy is built around “seamless Human + AI augmentation,” where intelligent agents become a core foundation for modern digital services, leading end-to-end enterprise workflows. This means roles evolve to collaborate with AI, rather than being replaced by it.
- Building AI-Native Expertise: We are fostering an “AI-Native Service provider and Agent builders of the world” workforce. This involves:
- Developing Intelligent Architecture Skills: Advancing capabilities to design and deliver “future-proof, intelligent architectures” that blend rule-based logic from legacy systems with the reason-based logic of Agentic AI.
- Deepening Trustworthy AI Capabilities: Building expertise in responsible, reliable, observable, and optimal AI (R2O2.ai). This includes skills in AI resilience, performance management, data drift detection, cost optimization, AI observability, security, governance, and agent lifecycle management, which are operationalized through HARC for AI.
- Enhancing IT/OT Integration: Cultivating “edge-to-core expertise” to manage data and integrate AI solutions across diverse technology landscapes, bridging IT and Operational Technology (OT) systems.
- Domain-Specific Agent Building: Scaling expertise in developing and deploying HARC.agents for key domains like Industrial, Security, Analytical, Operations, Engineering, and Cloud.
- Rethinking Job Roles and Responsibilities: For roles impacted by Agentic AI automation, such as invoice categorization (where we’ve seen a 70% reduction in human operators), responsibilities are shifting from routine execution to higher-value tasks like oversight, strategic management, and handling exceptions.
- Establishing Structured Innovation: The Center for Architecture and AI (CAAI) serves as a cross-functional, global working group to deeply and systematically embed AI into next-generation architectures, codifying knowledge into reusable AI assets and deepening capabilities across practices.
CIO&Leader: The report highlights potential productivity gains of over 40%. How are you measuring and realizing productivity improvements through AI in your business today?
Balasubramanian: While the 40% productivity gains is a broad view per the report, we are actively measuring and realizing productivity improvements through AI, particularly Agentic AI, by focusing on quantifiable business outcomes and demonstrable value. We make it use-case specific, for e.g.
- Reducing Due-Diligence Time: Agentic AI has reduced deal due-diligence time by 60% for a large private equity firm
- Automating Back-Office Processes: For a leading provider of sustainable packaging, paper products, and recycling services, an agentic process analyzes, validates, and categorizes large volume invoices, leading to:
- Increased accuracy to 90% from 65%
- Significant reduction in operational costs.
- Reducing downtime through predictive maintenance
- Finally, in the software engineering space, we see anywhere between 15 – 30% productivity improvement which we continue to measure using our Agentic POD framework.
CIO&Leader: As agentic AI systems gain autonomy in decision-making, how are you approaching governance, ethics, and transparency to ensure responsible AI deployment?
Balasubramanian: To ensure responsible AI deployment as Agentic AI systems gain autonomy, we are approaching governance, ethics, and transparency through several key initiatives:
- R2O2.ai Framework: We utilize and advocate for our R2O2.ai framework (Responsible, Reliable, Observable, and Optimal AI). This framework directly addresses the need for trustworthy AI, fostering confidence and accelerating enterprise-wide adoption by mitigating risks. It guides responsible AI development, deployment, and operations.
- Operationalizing with HARC for AI: The HARC for AI service operationalizes the R2O2.ai framework, bringing AI observability, lifecycle management, and performance tuning into real-world use. Its capabilities include:
- AI Resilience & Performance Management: Ensuring stable and available AI applications through monitoring and recovery strategies.
- Data Drift Detection: Tracking and correcting shifts in data patterns to maintain AI accuracy.
- Security & Governance: Protecting systems from attacks and enforcing compliance through automated controls.
- Agent Lifecycle Management: Supporting AI agents from deployment through continuous improvement.
- Robust Data Governance and Security: We are implementing comprehensive data governance policies to manage data privacy, security, and compliance, including defining data access controls, retention policies, and audit trails, which is essential as autonomous agents rely heavily on data.
- Building Intelligent Architectures: Architectural considerations are crucial, incorporating robust data governance, security, and frameworks like R2O2.ai to build, scale, and operate trustworthy AI solutions. This blends rule-based logic with reasoning-based logic for future-proofing.