Agentic AI is not a magic wand, says Arundhati Bhattacharya, president and CEO, South Asia, Salesforce
As he collects his papers after a meeting, Arundhati Bhattacharya—President and CEO of Salesforce South Asia—is occasionally approached by an attendee who wants to share a comment. A conversation begins and the person says something significant that would not be possible in a virtual conference where everyone quickly logs off.
Summary: While AI agents have captured the attention of technologists and leaders alike, an unsolicited conversation between two human beings can lead to insights that become the most important part of a business meeting.
Speaking to The Hindu at the Dreamforce enterprise technology conference in San Francisco, Arundhati Bhattacharya, president and CEO of Salesforce South Asia, discussed opportunities for young South Asian tech professionals, the evolving workplace culture in India, the nuances of deploying AI agents, cyber crimes in the banking sector and Salesforce’s ambitions in the country.
South Asians have a lot to offer the world as technologists, designers, scalers, builders and leaders like you. But many are hampered by very human biases that affect visa applications and job relocation opportunities. What is your advice to these young tech professionals right now? And how do you as a leader navigate this very problematic relationship between the US government and a South Asian workforce that is viewed with hostility?
I would like to say that we have had no such difficulties within the organization. Wherever there has been a demand for talent, we have it from any part of the world where it is available. Otherwise, a company like ours wouldn’t grow the way we did.
Overall, you have to remember that geopolitics will affect certain things at certain times. But still you must also remember that we have one of the youngest nations. And so if people are needed – no matter what kind of politics gets in the way – it will be temporary.
So I really don’t believe it’s something that will become a permanent barrier. But having said that, if we want to stay in technology, we need to make sure we have the right skills in the people who will stay there. So we need to improve our skills.
Today we know that most of the basic coding can be done by the AI agents we currently have. So vibration coding is very much there; our own people do a lot of vibration coding. But when it comes to really contextualizing it, when it comes to really fitting it into the architecture of what you already have—because no one is without anything—you always have to go and fit it into the existing way things work. Whenever you look at how you also make sure that the compliance is completely met, the regulatory requirements are met – whenever you need or do all of that, you still need people to do it.
You have to understand that things can happen from anywhere. At least technology has given us that freedom.
We are currently seeing many tech companies bringing people back into the office and ending remote work, with the culture being created in a cubicle environment. What is your take on that, especially in the context of South Asia?
We expect our workers to come three times a week. Because we think there should be a healthy mix. Working remotely gives you certain advantages, especially for people who have children or elderly people at home. But on the other hand, you also need to have an environment where you communicate with colleagues. Because not everything you learn only from your trainings and only from practice. Sharing experiences is a very good way to improve. And that happens when you’re in an office environment, in an office environment. It’s not so much about the cabins. These are more times when you get together. You’ve been to meetings and you know that sometimes the most important things are said in meetings only after the meeting is over.
So to that extent, it’s important for people to get into an office environment once in a while. But it’s also true that working remotely allows us to tap into talent that we might not otherwise be able to tap into. Women in the workforce, for example: many of them, given flexibility and mobility, would not leave the workforce. Maybe they would use these facilities for 2 years, 3 years until they get stable, before they put their kids in school, until their elders get better at home. That kind of flexibility will ensure that women don’t leave the workforce.
It will ensure that if the talent is sitting in another city, you will still be able to use it because you are willing to develop it.
We hear a lot about AI-assisted cyber attacks, building resilience and cyber defense strategies. Could you walk me through how these conversations are going in South Asia and India?
These conversations are happening everywhere, across. In India specifically, I will tell you: one of the things that I think we need to do very urgently is a lot more consumer education. Because, you know, your chain is only as strong as the weakest link. And today, the weakest link is the consumer. I know people who have shared their OTP 11 times. . 0.11 times before he called me and said “I think my money is disappearing.” They are people who have held high positions in organizations.
Agent AI is not widespread enough in these sectors now to have problems with cyber attacks and such. We are still at the very beginning of accepting these things. We haven’t even seen what they can actually deliver and do. Talking about cyber attacks carried out by agent means probably means putting the cart before the horse.
Let’s embrace it first and see where it goes. Do it with the right kind of railing and do it based on what you require. Don’t ask agents to do everything. Ask agents specifically to do “this and nothing more.”
Make sure all handrails are in place. Make sure the model is working in the harness, and the harness is something we talk about all the time because it sets the context. It gives you an idea of what that particular organization wants, what it does, what its vision is, where it needs to go. So, once you plug it in, you know, then you can really use its strengths in the right way. Imagine a horse in a harness. He can actually pull a carriage. Now you let it run its course. . .goes and destroys everything around.
What is the biggest misconception about agent AI that you clear up a lot?
The biggest misunderstanding is expectations from agent AI. It’s not a magic wand. You can’t wave your hand and solve everything. It has specific functions. You have to really understand what you want to do with it and then put it to work. So it’s a lot of power, but you have to temper it and use it for what you really need, rather than just putting it anywhere and everywhere. Lots of people make pilots. The pilots really don’t prove themselves. You have to find what really grips you. Find where the shoe is pinching and then try to fix that particular problem. And then you realize the value of what you are getting.
What is Salesforce’s relationship with the public sector in India and how is it shaping up?
So it takes time because if you want to get into the public sector business, you need a lot of certifications, you need registrations, you also need to make your case as to why you should be a company that should also be able to be involved in providing solutions.
But it’s starting to happen. We’re working with several ministries, we’re working with several states, and I believe there’s a lot we can do to help.