India must move beyond AI pilots: Arundhati Bhattacharya, President & CEO, Salesforce, South Asia
For more than four decades, Arundhati Bhattacharya has witnessed multiple waves of technological disruption reshape industries, institutions, and business models. From leading the digital transformation of the 200-year-old State Bank of India to now steering Salesforce’s South Asia business during the rise of generative and agentic AI, Bhattacharya has consistently operated at the intersection of scale, leadership, and transformation.
As the first woman chairperson of SBI, she drove one of the largest banking technology transformations in the country, modernised operations across thousands of branches, and executed the merger of SBI with its associate banks with minimal disruption.
Today, as the President & CEO of Salesforce, South Asia, she is once again watching enterprises navigate another defining shift, one that she believes could fundamentally alter how businesses operate, compete, and grow. In an exclusive interaction with Express Computer, Bhattacharya speaks candidly about India’s AI readiness, why enterprises must move beyond endless experimentation, the foundational role of data and workflows in AI success, and why the next phase of enterprise transformation will belong to organisations that operationalise AI rather than merely discuss it.
India’s AI race is now about execution, not curiosity
Bhattacharya believes India stands at a critical inflection point where AI adoption can no longer remain limited to boardroom ambition or innovation labs. According to her, enterprises now need to see AI as a strategic business imperative tied directly to productivity, competitiveness, and long-term growth. “I think it is important for Indian enterprises to really and truly take this up as a challenge that needs to be turned into an opportunity to improve on the efficiency and productivity of their organisations.”
She says Indian enterprises have already shown significant interest in AI adoption, but the pace of actual execution still remains slower than what global competitiveness now demands. “Is there interest? There’s a lot of interest. Is there a commitment that they will, you know, work on this? Yes, there is a commitment. Are we seeing a lot of work on the ground? We are seeing some amount of work, but we would like to see all of this proceeding much faster. Because today, the pace is becoming very, very important for us.”
For Bhattacharya, the global AI race is no longer theoretical. Enterprises that fail to operationalise AI quickly risk losing strategic ground. “The quicker we are able to improve on the pace, the better we will be positioned for being more competitive globally, and that is something I think that India wants to do and needs to do. That needs to stay competitive globally.”
The biggest AI mistake enterprises are making today
One of the strongest themes emerging from Bhattacharya’s interaction is her criticism of the way many enterprises continue approaching AI adoption through fragmented pilots and isolated experimentation.
According to her, organisations often hesitate because they remain uncertain about costs, returns, and business outcomes. “There is some amount of scepticism as to whether this is something that people really want to try out or not. There is also a question of cost, if I put in so much money, am I going to get returns that will justify this money being spent?”
This uncertainty, she says, pushes organisations into creating multiple pilots that rarely translate into production-scale value. “As a result of that, what happens is they go ahead, and then they start creating pilots. But pilots, we have to remember, are not full-fledged production. So, a pilot may not really show you the value.”
Bhattacharya also believes enterprises need to fundamentally rethink this approach. Rather than experimenting endlessly across disconnected use cases, companies should focus on solving a few high-impact business challenges properly and at scale. “Our recommendation always is that, instead of doing a pilot, take one or two really important use cases. Use cases which you believe will actually showcase the value to you. Try them out and then see whether you want to try more.”
According to her, the enterprises that succeed in the AI era will not necessarily be the ones running the largest number of pilots but the ones capable of moving quickly from experimentation to execution.
Agentic AI will fail without strong data and workflow foundations
Bhattacharya reminds that one of the most overlooked aspects of AI adoption today is enterprise preparedness. Many organisations, she says, are trying to deploy intelligent AI agents on top of fragmented data systems and broken workflows.
According to her, the first foundational layer is enterprise data. The second layer revolves around workflow automation and operational structure.
Only after these two layers are stabilised can enterprises begin extracting meaningful value from agentic AI systems. “You need to go step by step in order to ensure that the agentic piece, once it rides on a properly created data foundation and an automated workflow space, then it works very well.”
Bhattacharya’s larger message is that AI success is not just about buying technology. It is about organisational readiness “We need to be very thoughtful about the way we go about it, the way we implement it. Simply running pilots doesn’t really get you there, because pilots will cause many more questions to come up than actual answers because it’s not in production.”
Salesforce sees India as a long-term strategic AI market
Beyond enterprise AI adoption, Bhattacharya also reflected on Salesforce’s growing India presence and the role India plays within the company’s global engineering and innovation ecosystem.
Despite global conversations around AI-led hiring slowdowns, she clarifies that Salesforce continues to expand its talent base depending on evolving business requirements.
“For the last six years, there’s not been a single quarter when we have not hired. So, hiring continues depending upon our requirements. And as the requirements change, the capabilities that we are looking for may change.”
She points out that Salesforce’s India operations have scaled significantly over the years. “We started with 2500 people, so we have grown. To that extent, yes, I am satisfied by the way the operations have grown in this country.”
Bhattacharya also stresses that Salesforce’s engineering model is fundamentally global rather than region-specific. “All of our products and engineering are global in nature. It is not local in nature. Wherever the teams are placed and a lot of teams are placed in India as well, they serve the global mandate.”
The global product roadmaps are centrally aligned, while innovation and execution happen across distributed engineering teams worldwide. She also cites upcoming multilingual voice capabilities as an example of how global platforms are increasingly becoming locally contextual.
Why unified customer intelligence becomes critical in the AI era
Bhattacharya repeatedly emphasises on one core idea during the interaction, the importance of unified enterprise data.
According to her, Salesforce’s Customer 360 approach becomes even more relevant in the age of agentic AI because intelligent systems depend heavily on integrated customer intelligence. “We believe that we are a very relevant technology for anybody wishing to create unique customer experiences. Because we provide the unification of the data regarding a customer.”
She explains that Salesforce’s CRM architecture is designed to create a unified customer record across every business interaction. “The customer 360 golden record, so to speak, is what we create in our CRM, and over and above the customer record, we create the sales opportunities, the service treatments, the marketing, the commerce, all of that, and the analytics.”
For her, enterprises that successfully unify customer data will ultimately build stronger AI-driven customer experiences.
AI is changing pricing models, business models, and enterprise operations
She also acknowledges that AI is forcing enterprise technology companies themselves to rethink traditional operating structures.
Speaking about Salesforce’s Agentforce strategy, she explains how AI consumption patterns are reshaping pricing models as well.
“The licence-based model doesn’t work everywhere. Today, we are also evolving the pricing packages.”
According to her, enterprises increasingly require flexibility between seat-based and usage-based AI pricing structures. “A consumption model sometimes is more conducive than merely a per-seat one. But we have also seen companies where we have gone with the consumption model, and they prefer the per-seat.”
As AI evolves, Salesforce is therefore redesigning not only products but also go-to-market models and customer engagement structures.
“We are trying to ensure that we have enough numbers of these pricing packages for people to be able to accept whatever they believe meets their needs. The way that we used to do things earlier, we are also changing not only the products but also the pricing models.
Also, the go-to market, all of that, is changing because the technology is such that it demands that kind of change.”
Enterprises that operationalise AI fastest will define the next decade
Bhattacharya’s broader message throughout the interaction is ultimately about urgency.
India, she believes, has both the market opportunity and technology capability to become a major AI-driven economy. But success will depend on whether enterprises can move beyond fragmented pilots and begin embedding AI deeply into operational workflows, customer engagement, and enterprise decision-making.
For her, the next phase of enterprise transformation will not be defined by who experiments with AI first but by who operationalises it fastest, most responsibly, and at scale.
