Enterprise Data Platforms Unlock True AI ROI When Architecture Leads And The Right Tools Follow

Enterprise data platforms are becoming the active operating layer of the business, powering decisions in real time rather than sitting as passive warehouses. For many CIOs, making this transformation requires a disciplined approach that prioritizes architecture and business outcomes over a reactive, tool-first rush into AI.

Sandeep Bansal is Chief Information Officer at A-One Steels India Limited and a Board Member of the World AI Governance Foundation. A CIO100 Awardee, he has built a career on turning sprawling IT environments into models of efficiency. At Manipal Academy of Higher Education, he led a large-scale application consolidation, reducing 180 systems to just 15 core enterprise platforms. His perspective offers a pragmatic counterpoint to the hype surrounding AI.

"Most of what I see in AI is overhyped right now because people treat it like a magician that will do everything,” Bansal said. “It doesn’t work that way. Before introducing AI, you must know exactly what you want to achieve, where it fits in your architecture, and how governance will control it." It is a philosophy rooted in a conviction he has carried across his career: transformation is not about velocity, it is about the outcome.

For Bansal, building a proper foundation means rethinking data's role entirely. Rather than sitting at rest, data must fly across a cohesive, integrated ecosystem and deliver answers in real time. That requires real-time data architecture supported by a governance-first mindset, one that builds data trust by embedding safeguards from the beginning.

• Human in the loop: When critical systems trigger ambiguous alerts, the cost of a wrong automated decision can outweigh the efficiency gain. "If my monitoring shows our disaster recovery site is down, I would not allow AI to automatically initiate the DR process. The alert could be a false positive from a single component. AI cannot make that judgment call," Bansal said. "It requires a human to investigate and initiate."

• Garbage in, garbage out: The same discipline applies to how AI is trained from the outset. "Training AI with ethical use in mind is like raising a child. You can give it all the right information, but if it ultimately doesn't behave the way you want, it means your fundamental training was not right," Bansal said.

Bansal's philosophy translates into an actionable CIO playbook for engineering an AI-ready enterprise. It is a direct response to the familiar challenge of tool sprawl, where different business units want their own solutions. He said any solution must first be designed to align with the organization's vision and desired business outcomes, not just added as another tool. While many organizations still lack the AI maturity and discipline to calculate returns, his method aims to provide a clear path by focusing on a unified operating system for value and deploying AI agents effectively.

• Architecture first, tech second: Every tool request from a business unit begins the same way for Bansal, with a question about need and purpose before any technology decision is made. "When a business user comes to me for a new application, my process begins by asking what is needed and why. Only then do I design the architecture before deciding which technology we are going to use," Bansal said.

• Benefits over features: His approach to true ROI is a two-part formula that balances cost efficiency with adoption, most recently delivering a 42% cost reduction alongside a 99% adoption rate. "An application's success isn't that you implemented it. It's whether users adopt it. That is the value I want to drive," Bansal said.

His perspective encourages CIOs to become long-term planners, moving beyond the day-to-day role of a tactical firefighter. It is a mindset, he said, that helps separate the winners and losers in the AI era and defines modern strategic leadership. He shared a final story about being asked by his CEO to design a technology vision not for the next quarter, but for the next decade. He distilled that thinking into five questions every CIO should be able to answer: what value will you drive, what methodology will you use, what risks do you foresee, what metrics define success, and where does the organization's vision point?

He believes CIOs should focus on presenting the board with a long-term strategic picture, shifting the focus away from firefighting toward becoming a revenue-generating partner for the business. "As a leader, my responsibility is to show the board a strategic picture, not just react because AI has come with an impulse to implement it today. I will implement it next year because I don't need it today. By that time, it will get mature as well," Bansal concluded.