“AI cannot fix what incomplete data creates”: Deepu Chacko on India’s agentic AI challenge

As Indian businesses ramp up their artificial intelligence (AI) initiatives, a major hurdle lies not just in computational capabilities or the complexity of models, but in the quality of data itself. Deepu Chacko, Vice-President of Solution Engineering at Salesforce India, emphasizes that inconsistent and unreliable data is a significant barrier to achieving AI goals. "AI cannot remedy the problems created by incomplete data. For India to fully harness the potential of agentic AI, it is imperative that leaders treat data as a vital resource—unified, governed, and contextually relevant," Chacko remarked in a recent interview. The issue of fragmented data systems is prevalent; a decade ago, Indian enterprises were primarily focused on digitalization. Nowadays, they produce vast amounts of data, from customer interactions to backend transactions. However, much of this information remains isolated. Chacko stated, "Orders may reside in ERP systems while sales data is housed in Salesforce, and other engagement metrics are scattered across different applications." This fragmentation poses a risk to AI efficacy, as outcomes diminish when customer information is spread across disjointed systems. Highlighting the importance of data integrity, Chacko pointed to a Salesforce report indicating that 89% of data and analytics leaders believe that the quality of AI outputs is directly tied to data inputs. Alarmingly, these same leaders estimate that around 25% of their organizational data is unreliable. Chacko bluntly stated, "Bad data is equal to bad AI." Over the years, India's swift digital transformation has enhanced the richness of customer data, evolving from simple identifiers like mobile numbers to comprehensive profiles that include email addresses and personal preferences. This evolution necessitates stronger data governance; companies that neglect to modernize their data infrastructure risk implementing AI on shaky foundations. Amid rising concerns regarding AI hallucinations, Chacko warns that organizations may be concentrating on the wrong aspects. While prompt engineering—directing AI systems on how to behave—plays a role, it is not enough on its own. "Context engineering," which involves utilizing existing enterprise data, is crucial. An AI assistant that understands a customer's history, spending habits, and previous interactions can tailor communications with greater accuracy. Chacko stressed the need for robust data access and the ability to analyze AI interactions, asking, "Why did the AI respond that way? What context did it have?" Looking forward, Chacko predicts that the next evolution will be agentic AI—autonomous systems capable of interacting with other AI agents. He foresees a future where personal AI agents manage tasks such as travel bookings and negotiations directly with brand representatives. An "agentic enterprise," according to Chacko, must adapt its systems and workflows to facilitate these AI-to-AI interactions. "Agentic AI isn’t just a technological advancement; it's a revolutionary change. These agents will handle routine tasks, allowing humans to dedicate their efforts to creativity and relationship-building," he explained. In terms of workforce dynamics, Chacko noted that the narrative of AI disproportionately benefiting younger, tech-savvy individuals is misleading. He observed no significant age-related displacement, as automation allows seasoned professionals to shift their focus from repetitive tasks to strategic relationship management. "Senior leaders can gain an advantage if they have access to integrated data. Meaningful conversations with customers hinge on context, which in turn relies on cohesive systems," he concluded.