A World Bank Exec's Vision For An Agentic-First Workplace

*The views and opinions expressed by Sheila Jagannathan are her own and do not necessarily represent those of any former or current employers.

Most companies have fallen into the AI trap: while 80% report using it, far fewer see significant ROI. The market might be flush with tools, but the meaningful results are scarce. Now, some experts say the most viable solution is a radical re-engineering of the entire system.

For an expert's take, we spoke with Sheila Jagannathan, Global Head of Capacity Building and Digital Learning at The World Bank. An internationally recognized thought leader and author on digital learning with over 35 years of experience leading human capital development and transformation, Jagannathan offers a pragmatic view into the widening gap between technology's immense potential and the limitations of our current reality. Before organizations can succeed with AI, she said, leaders must stop treating it like a plug-and-play solution.

"My biggest piece of advice is don’t squeeze it into your old processes. Instead, design your workflows from the ground up with agentic AI at the center. That's the way to do it right," Jagannathan said. This bolt-on approach is precisely why so many AI initiatives are failing to deliver, she explained. For her, it's a clear sign that the current strategy is flawed.

• From insight to inaction: The problem is a fundamental misunderstanding of AI's role, according to Jagannathan. "Typically, many people think of AI as merely a specialized tool, be it a chatbot or an analytics engine. So, it's often deployed quietly, with minimum oversight, and siloed by function."

This type of approach usually stops at passive insights, Jagannathan explained. But further complicating the matter is agentic AI. Agents can transform passive insights into active execution, but only if the conditions are right. Most of the time, without first rethinking the entire workflow, AI can't take action.

• Onboarding digital teammates: For Jagannathan, the solution is a mindset shift. "When you bring in a new teammate or a new hire, you don't just throw them in without any guidance. You give them a role, a terms of reference, permissions, performance goals, and regular check-ins. Now, leaders can onboard agents the same way: gradually rolling them out, monitoring how they perform, and making sure they align with the organization's mission."

• From operator to co-creator: Naturally, the transition from rigid org chart to fluid "orchestration graph" also requires some new skills, Jagannathan said. "It's a shift from simply executing tasks to orchestrating, supervising, and providing strategic oversight. In all of this, trust and transparency are crucial. The bottom line is that agentic AI ushers in a fundamental shift from AI that passively responds to human prompts to AI that actively orchestrates, anticipates, and executes with design and intent."

Tying technology directly to tangible value creation is the ultimate goal, Jagannathan advised. But as the nature of AI's contribution changes, so too must the metrics for success. "It's not just about dollars. It's about faster learning, more personalized feedback, stronger collaboration, higher morale, more trust, and better decision-making." Until humans fundamentally evolve our relationship with technology, the future of work will be an intangible solution. "The real leap isn't when machines think for us," she concluded. "The real leap is when the machines think with us."