BT Exclusive: ‘Human-in-the-Loop is AI’s biggest safeguard’: Arundhati Bhattacharya on the agentic era

At a time when players like Anthropic are rapidly pushing the boundaries of what AI systems can do, moving from copilots to autonomous agents; AI is no longer just a passive tool but an active decision-maker. The conversation, therefore, is shifting beyond adoption to questions of control, accountability, and intent. At the Trailblazing Women Summit 2026 hosted by Salesforce, that shift took centre stage, with leaders underlining a critical truth that in the agentic AI era, humans are not being replaced, they are being repositioned.

Speaking exclusively to Business Today, Arundhati Bhattacharya, President and CEO, Salesforce South Asia says that the real challenge lies not in the sophistication of the technology but in how organisations deploy it. Agentic AI, defined by its ability to take actions, trigger workflows, and independently resolve tasks, is already reshaping enterprise operations.

“You cannot drop intelligence into a vacuum,” she said, cautioning that without unified and governed data, even the most advanced AI systems risk becoming performative rather than productive.

For Indian businesses, often characterised by ambition and scale, this marks a pivotal moment. The gap between experimentation and enterprise-wide value creation, Bhattacharya argued, stems from treating AI as a standalone tool instead of embedding it into the organisation’s core architecture. The companies that bridge this gap will define the next decade.

Yet, as AI systems become more autonomous, the role of humans becomes more—not less—critical. The idea of “human-in-the-loop” is evolving beyond a compliance checkpoint into something far more nuanced. It is about designing systems with clear boundaries. Where AI acts, where it recommends, and where human judgment is non-negotiable.

Bhattacharya, drawing from decades of leadership experience, emphasised that judgment shaped by context, ethics, and lived experience cannot be replicated. “AI doesn’t diminish human judgment, it elevates it,” she noted. In practice, this means freeing humans from repetitive, transactional tasks while placing them firmly at the centre of high-stakes decisions involving trust, relationships, and long-term impact.

This shift is also redefining leadership itself. In an agentic world, speed and innovation can no longer come at the cost of governance. Leaders must balance competitive urgency with ethical responsibility, ensuring their workforce is prepared, their customers protected, and their systems resilient.

“We are at an inflection point, not just for technology, but for leadership itself,” Bhattacharya said on the sidelines of the summit which saw participation from over 380+ in-person attendees and 3,700 live virtual participants, from across industries. The real test, she suggested, is whether organisations can move beyond reacting to AI disruption and instead shape it with intention.

In the end, the agentic AI era may be defined not by how intelligent machines become, but by how wisely humans choose to guide them.