AI frontier: Acquiring people's knowledge before they retire

There are engineers at steel plants who can tell a bearing is about to fail simply by listening to the machine. Others know exactly how different grades of iron ore should be blended for maximum efficiency because they have spent decades watching furnaces behave under different conditions. Much of that expertise has never been written down.
Now, some of India’s biggest manufacturers are racing to capture that knowledge before it disappears. Rather than viewing AI purely as a tool to automate work, companies including JSW Group, Sun Pharmaceutical Industries and Parle Products are increasingly using it to preserve decades of institutional memory – building AI systems that can answer questions the way their most experienced employees would.
Harish Mehra, group CIO at JSW Group, told us during a webinar recently that the company is developing AI-powered knowledge graphs that combine maintenance records, engineering documentation, equipment data and the expertise of veteran employees. The idea is that a new engineer should be able to ask why a machine is behaving a certain way and receive an answer built on years of operational experience.
“It isn’t enough to feed documents into an AI model,” Mehra said. “You need the people who understand the process, the equipment manufacturers who know where failures typically occur, and technology teams that can connect all of that. That’s how you build something that lasts.”
At Sun Pharma, AI is not confined to isolated applications in procurement or finance. Instead, the company has built an internal AI assistant that can search infor mation flowing from around 50 enterprise systems, allowing managers to query manufacturing, supply chain, quality and commercial data through natural language.



“The breakthrough wasn’t deploying another chatbot,” said Dheeraj Sinha, CIO at Sun Pharma. “It was connecting knowledge across the organisation.”
About 20,000 managers now have access to the system, while the company’s 45,000 employees have generated around 7.5 million AI prompts in six months.
Parle Products is applying the same principle to one of India’s largest consumer distribution networks. The biscuit maker has developed an internal AI assistant called ChatG that allows sales teams to ask business questions conversationally instead of waiting for reports or dashboards.
“The real change isn’t that people have access to AI,” said Sanjay Joshi, CTO at Parle Products. “The change is that they can access years of business knowledge instantly. Someone can ask about sales performance, product movement or distributor behaviour in plain language and get an answer immediately.”
The company is also using AI to rethink how products move through its distribution network. Historically, distributors were supplied largely on the basis of past buying patterns. AI now combines historical sales with external market signals to recommend product mixes tailored to individual wholesalers and locations.