IBM executive: Hybrid infrastructure is the foundation for scaling AI agents
Enterprise AI is moving beyond automating routine tasks to deploying autonomous agents that manage complex workflows in real time. As organisations expand these systems, concerns about trust, governance, and data sovereignty are reshaping how companies build AI infrastructure.
Gaurav Agarwal, vice-president of Technology at IBM India & South Asia, said the shift requires rethinking how enterprises handle data, infrastructure, and control. The company is positioning its watsonx platform and hybrid cloud approach as the backbone for scaling AI agents responsibly.
Where AI agents differ from traditional automation
Traditional systems execute predefined tasks within isolated processes. AI agents operate differently - they understand context, make decisions, and act autonomously within business workflows.
"This shift to workflow-centric operations is critical," Agarwal said, "enabling organisations to reduce friction, accelerate decision-making, and deliver smarter outcomes through continuous learning and human-AI collaboration."
Scaling from pilot to production
Moving AI agents from proof-of-concept to enterprise scale requires three things: selecting the right framework aligned to business goals, balancing sophistication with simplicity, and maintaining strong data privacy controls.
Agarwal said overly complex systems slow adoption. "Ease of use accelerates value," he said. Testing for performance and accuracy before wider rollout prevents reliability problems as deployment grows.
Orchestration platforms matter here. Agarwal said IBM watsonx brings together orchestration, governance, and performance management to help organisations scale agents across the business.
Data architecture determines outcomes
Clean, governed data available in real time is the foundation for AI success. Many organisations still have data fragmented across systems, arriving too late to influence decisions.
IBM's acquisition of Confluent addresses this gap by enabling real-time data streaming. The platform embeds lineage tracking, policy enforcement, and quality controls so AI systems can act on continuously updated enterprise data.
Infrastructure must be hybrid by design
Legacy systems built for a different era lack the modularity modern enterprises need for agentic AI. An IBM study found that while 70% of executives say hybrid strategies optimise cost and performance, only 42% are confident their current infrastructure can handle the data and compute demands of advanced AI models.
A hybrid-by-design approach deliberately integrates on-premises and multi-cloud environments. "In the race for AI leadership, competitive advantage will belong not to the flashiest AI tool, but to the resilient, invisible architecture that supports it," Agarwal said.
Real business results already visible
IBM embedded AI across its own enterprise workflows and achieved $4.5 billion in productivity gains globally. The company's AskHR agent autonomously resolves 94% of employee queries, reducing support tickets by 75% and cutting operating budget by 40%.
In IT operations, AI agents have reduced support tickets by 56%, with systems resolving nearly 86% of queries. Finance teams are using agents for trading, compliance, reporting, and risk management. The biggest impact appears in financial reporting and accounting, where agents automate data collection, validation, and disclosure while flagging risks.
On production floors, AI agents adjust machines in real time, trigger quality checks, and trace root causes across manufacturing processes.
Governing autonomous systems
As AI agents gain autonomy, governance becomes critical. Agarwal said digital sovereignty - ensuring data residency, access controls, and policy enforcement - must be embedded from the start, not added later.
Continuous compliance monitoring aligns AI operations with evolving regulations through real-time auditability and risk controls. "By integrating governance and elevating it to a strategic priority, organisations can build trust by ensuring accountability, security, and aligning with business and regulatory expectations," Agarwal said.
For management teams implementing agentic AI, understanding both the infrastructure requirements and governance framework is essential. Learn more about AI for Management and Generative AI and LLM foundations to support enterprise-scale deployments.
Your membership also unlocks:
