Enterprises layer AI onto existing workflows without redesigning them, expanding roles and increasing workloads rather than delivering on AI's productivity promise.
Anna Jibgashvili, Corporate VP at New York Life Insurance Company, says treating AI as a tool rather than a transformational technology is why enterprise workloads aren't shrinking.
She outlines a framework for defining which work belongs to machines and which requires human judgment, shifting roles from producing outputs to verifying AI-driven results.
Enterprise AI transformation is failing to deliver on its promise of productivity, but the obstacle is rarely the technology. Most organizations have deployed AI onto existing workflows without redesigning how work gets done or how decisions are made. Adoption is accelerating faster than organizations can absorb it, creating an AI readiness gap that no tool can close. The missing ingredient is leadership clarity: defining what work matters, what belongs to machines, and where human judgment remains irreplaceable.
Anna Jibgashvili is Corporate Vice President and Data Product Manager for AI and Data at New York Life Insurance Company, where she leads enterprise AI transformation strategy across one of the most heavily regulated sectors in financial services. She has held senior data and AI leadership roles at Guardian Life, Moody's Analytics, and S&P Global, accumulating more than fifteen years of experience building the operating foundations that allow AI to scale safely. Jibgashvili believes that most enterprises are deploying AI in a way that is quietly compounding the problem rather than solving it.
"We're introducing AI as a tool, but it's not a tool. It's a transformational technology," says Jibgashvili. "When you treat it like a tool, you automate a use case but never redesign the work around it. That's why workloads aren't actually shrinking." It is a pattern Jibgashvili sees repeating across enterprises of every size and sector.
This "tool-first" mindset paradoxically increases workloads and stretches existing roles beyond their original scope, often preventing true scalability. Jibgashvili calls this phenomenon the "ElasticPM," where a product manager's role suddenly expands far beyond strategy. It adds difficult new responsibilities like agent design, data sourcing, model governance, and user adoption, reflecting the changing nature of product development. This flawed approach also encourages a focus on the wrong things.
Wrong scorecard: "Unfortunately, what we are doing right now is trying to scale by implementing as many AI agents as possible, as if that is the measure of success," Jibgashvili notes. For organizations still measuring AI progress by deployment volume, the real costs, including overextended roles and stalled productivity, remain invisible.
Fear and failing: This pressure to deploy at scale creates consequences inside organizations, both for the people doing the work and for how leaders measure progress. "That fear of AI comes from not having a clear picture from leadership about what we are doing," says Jibgashvili. "We are not replacing humans. We are empowering them," she adds. "We empower them by redesigning roles to show that their job is no longer just gathering data, but analyzing it to find new business insights."
Business over buzz: The solution, she says, has less to do with technology fluency than most leaders assume. "You might be surprised by this answer, but I don't think tech-savvy leadership is the solution here," says Jibgashvili. "To truly become an AI enterprise, you need leaders who understand the business: how you make revenue, what the processes are, who the approvers are, and where your bottlenecks are. That tribal knowledge is what makes the enterprise work."
Her focus on business fundamentals sets up her core, actionable framework. A leader's primary job, Jibgashvili says, is to codify that "tribal knowledge" into what she calls an "enterprise digital identity." Once that identity is established, it becomes the basis for determining exactly where human judgment ends and machine execution begins.
Code your culture: "Leadership's responsibility is what I call creating an enterprise digital identity," says Jibgashvili. "You must define the process that you uniquely own as an entity so that it is digitally legible. Then, technology can work with it." Without that legibility, she says, AI implementations remain siloed, unable to scale or communicate across the enterprise.
Producer to proofer: With this framework in place, roles can be redefined with purpose. Redefining roles effectively depends on strong human oversight in AI governance and clear accountability in AI workflows, a shift already reshaping the future of jobs. Every role, she adds, must be analyzed to determine what is "machine eligible" versus where human judgment is required. "You're no longer producing that report, but you are responsible for ensuring the proper filters are used and for verifying the outcome is accurate," Jibgashvili clarifies. "The ability to understand what an agent is producing is a completely different cognitive ability than simply creating a report."
From Jibgashvili's perspective, the truly irreplaceable skill leaders should focus on cultivating is human judgment rooted in historical context. While technical upskilling is a relatively straightforward challenge, it is this uniquely human cognitive power that provides lasting value. "What you cannot replace is the cognitive power to make the right decisions when AI makes mistakes," says Jibgashvili. "A digital identity doesn't capture how leadership handled a past emergency, a merger, or an economic event. These things are not documented. You still need a human to make those judgment calls."
That question of human judgment ultimately scales beyond the enterprise. Jibgashvili's concept of a digitalized organizational identity raises deeper questions about ownership in an AI-driven world, questions she sees as unresolved. Her concern connects to emerging debates around AI and data sovereignty and who ultimately controls what organizations digitize about themselves. "As an enterprise, you have to digitalize yourself for a proper AI transformation," she says. "Then the question becomes: who owns that identity?"
