Corporate AI initiatives are driven by marketing hype and short-term goals, with true customer-facing ROI likely five to ten years away in most industries.
Shreepada Rao, VP of Data Governance Management at State Street, argues the core problem is not technology but a corporate culture that fails to give teams a safety net to experiment, fail fast, and iterate on end-to-end solutions.
Executives must stop chasing new technology and instead deeply analyze current business processes to understand exactly where AI can deliver structural, long-term value.
Most enterprise AI programs never solve the problem they set out to solve. Teams optimize a narrow process for a single department, declare a quick win, and move on. The result is a growing collection of siloed solutions that address perhaps five percent of a larger organizational challenge while delivering little measurable value to the actual customer.
Shreepada Rao, Vice President of Data Governance Management at State Street, brings experience across data science, AI adoption, and analytics leadership at organizations including AGCO Corporation and Infosys. His work spans customer analytics, predictive modeling, and the organizational challenges of scaling data-driven decision-making. From his perspective, the fundamental barrier to enterprise AI success is not the technology. It is a corporate culture that pressures teams to prove value too quickly, producing fragmented results that fail the broader business.
"It's easy to add technology, but it's a huge effort to know where and how it should be implemented to make a real difference," says Rao. The pattern he describes is familiar. Initiatives launch with urgency, target a narrow use case, and try to demonstrate ROI before the team has mapped how the solution connects to the organization's end-to-end workflow. That pressure, Rao argues, produces brittle systems that solve small problems but miss the larger opportunity. Meanwhile, boards continue to push for faster returns on AI investments, compounding the tendency toward shallow implementations.
A safety net: For Rao, the single most important thing an organization can do is give teams permission to fail. "The best benefit comes when companies provide the opportunity to fail fast and get feedback to the developing teams," he says. "Unless we fail, we remain stuck in the same loop of trying to solve the same problem." Without that psychological safety, teams default to cautious, narrow solutions that serve their own department but add no structural value to the organization.
Culture, not technology: Rao is clear that the bottleneck is not technical capability. Computation is cheaper, skill levels are higher, and companies are more open to new tools than at any point in the past two decades. "It's not a technology problem," he says. "The skill level is quite high compared to now, to twenty years ago. We have done so much better in terms of all of these." What is missing is the organizational willingness to let teams experiment at scale. Large, well-established companies in particular tend to move slowly, preferring to acquire a company or hire consultants rather than develop capabilities in-house.
Hype before harvest: Rao sees much of the current AI investment cycle as marketing rather than genuine value delivery. "All of these initiatives are more of a hype. The companies are putting it into the market rather than providing the actual benefit out of this technology at the moment," he says. Autonomous agricultural equipment, for instance, generates customer excitement years before reaching the market. The automotive industry followed the same trajectory, talking about autonomous driving a decade ago before consumers began seeing real products. The pattern, in Rao's view, is consistent: investment and marketing outpace genuine value in the majority of organizations, with actual customer benefit arriving five to ten years after the initial push.
Banking's regulatory constraint: In heavily regulated industries like banking, the dynamic is even more constrained. Rao points to financial services as a sector where scrutiny from central authorities limits the scope of AI adoption to back-end process optimization. "You cannot come up with new technology that would solve the problem drastically, but definitely, there is so much scope for optimization," he says. Banking operations no longer require customers to leave the house for basic transactions, a genuine improvement. But regulatory guardrails prevent the kind of radical innovation that other industries can at least aspire to.
The common thread across industries is an executive knowledge gap. Technology acquisition is straightforward. Understanding where to apply it requires a deep, honest assessment of current business processes and customer needs. Most leaders, Rao observes, skip that step.
For executives navigating this tension, Rao's advice is direct: stop leading with the technology. "The focus should be to understand their current business needs, or the customer needs," he says. "Business needs to optimize the current processes and provide additional products and services to the customer." Until that foundational work is done, he warns, AI programs will continue to deliver marketing wins and siloed efficiencies that never reach the people they are supposed to serve.
