We are probably the last generation to only work with humans: Salesforce's Arundhati Bhattacharya on the agentic future

New Delhi this week became the centre of gravity for the global AI conversation, as world leaders, tech CEOs, and policymakers converged at the India AI Impact Summit 2026. Amid the noise of announcements and panel debates, one question kept surfacing: what does the rise of AI agents actually mean for the people who work alongside them?

On the sidelines of the summit, TimesofIndia.com sat down with Arundhati Bhattacharya, President and CEO, Salesforce South Asia, and Srini Tallapragada, President and Chief Engineering Officer, Salesforce, to discuss the conundrum.

The idea of working alongside AI agents rather than just using AI tools is no longer a distant concept—it is, according to these leaders, the defining shift of this decade. And far from being a cause for anxiety, they see it as an opportunity to fundamentally reimagine what human work looks like.

"We are probably going to be the last generation of people that will only handle human interactions," said Bhattacharya. "Going forward, the next generations will handle both agents and humans. That's our view of the future."

It's a striking claim, but Arundhati Bhattacharya made it with quiet confidence—not as a warning, but as an inevitability that businesses need to start preparing for now.

Everyone becomes an agent manager

At the heart of Salesforce's pitch is a reframing of what jobs actually are. Every role, the company argues, is essentially a collection of tasks and skills—some routine, some requiring judgement, some requiring deep experience. AI agents, in their view, are not job replacements. They are task absorbers.

"Humans do what they are good at—judgement, experience, guiding the process," said Srini Tallapragada, President and Chief Engineering Officer, Salesforce. "What agents will continue to do is the routine, the regular. In some cases, where you have well-defined rules and standard operating procedures, the agent can handle that too. Increasingly, humans will focus on more and more complex situations."

The practical implication, Tallapragada said, is that every person in an organisation will eventually manage agents—much like how every employee today manages email or uses a spreadsheet. "If you are a people manager today, you will manage humans and agents. If you are an individual contributor today, you will always manage agents. Almost everybody will become an agent manager and a builder. That's how this will transition."

Bhattacharya put it even more directly: the enterprise of the future is an agentified enterprise. "The agentic enterprise is coming," she said. "And we are going to help our customers get there."

But first, trust

For all the optimism, Salesforce is acutely aware of why AI has largely stayed in pilot mode for the past three years. The worry is simple: what happens when you put this at scale? Will it hallucinate? Will it make the wrong call? Can it be manipulated? Will it behave ethically?

"Trust is our number-one value," said Tallapragada. "One of the first things we build, before building any application, is the trust layer. Because we knew from day one—if you cannot handle all of these cases, businesses won't adopt it."

That trust layer has two components: preventative and reactive. On the preventative side, Salesforce runs red teams that rigorously test products before they ship, and has an Office of Ethical and Humane Use that governs AI development. But the work doesn't stop at launch.

"When a customer is implementing it, we provide a lot of guardrails and auditing capabilities," Tallapragada explained. He gave a concrete example: a financial services company using an AI loan agent. "Six months later, the regulator asks—how do I know you didn't discriminate? You need to be able to show exactly how the agent reasoned, what it said, what its logic was. You need audit trails, governance, PII masking, data residency compliance, alignment with all local laws."

The approach seems to be working. Agentforce is currently Salesforce's fastest-growing product, with thousands of enterprise customers implementing it. Notably, Tallapragada pointed to what the company calls "refilling the tank"—a significant portion of Agentforce customers are returning to expand their deployments after initial success. "That wouldn't happen," he said, "if we hadn't built the right foundational structure."

The message from Salesforce, ultimately, is one of structured optimism: the agent era is not a disruption to be managed, but a transition to be designed—and the companies that build the right foundations now, they argue, will be the ones that lead it.