Beyond automation: How RJ Corp is building an AI-powered operating model

For a retailer dealing with a malfunctioning beverage cooler, the difference between a traditional service model and an AI-enabled one can be as simple as the time on the clock. Earlier, a problem reported at late evening or midnight meant waiting until the call centre reopened at 9 a.m. Today, an AI-powered multilingual voice bot can identify the retailer, diagnose the issue, offer a solution, raise a service ticket, or hand the call to a human agent.

At RJ Corp, that shift captures something much larger than automation. Under Group CIO Dr. Kamal Karnatak, technology is increasingly being used to move intelligence closer to the point of decision—whether that is a retailer, salesperson, distributor, store employee, recruiter or manager.

The objective is not technology for its own sake. It is to make the organisation faster, more responsive and more capable of making decisions at scale.

Karnatak’s philosophy is straightforward: the employee dealing directly with the customer should have the right data to make the right decision. He describes this as the “democratisation of decision-making.”

That idea provides a useful lens through which to understand RJ Corp’s AI journey.

From dashboards to decisions

RJ Corp’s relationship with data predates the current generative AI boom. The group had already been using analytics for forecasting and suggested orders before the current AI wave accelerated after 2022. What has changed is the breadth and accessibility of that intelligence.

Consider its beverage business. Around 1 million Visi coolers are deployed with retailers. The organisation’s AI voice bot can verify the retailer using the asset number, draw on a knowledge base to troubleshoot the issue, create a ticket when required, and escalate the interaction to a human when the bot cannot resolve it. The service is available in multiple languages, including Hindi, English, Telugu, and Tamil.

The important point is that the business case was not primarily about reducing call-centre costs. It was about changing the service promise from limited operating hours to 24×7 availability.

The same principle is being extended to decision-making. Instead of expecting managers to navigate static dashboards and then request additional information from IT or analysts, RJ Corp has created an internal ChatGPT-like interface, DISHA (Decision Support Intelligent Smart Help Assistant), and an AI CoE connected to its data.

A manager can ask why sales in Madhya Pradesh are declining, for example, and have the system analyse the available data. The information remains within the organisation’s environment.

That represents a significant shift: data is no longer simply something the organisation reports; it is becoming something employees can interrogate.

One backbone for very different businesses

The challenge is amplified by RJ Corp’s diversity.

Its portfolio spans beverages, QSR, retail, ice cream, healthcare and education, with businesses operating at very different levels of maturity and with very different workforce profiles. Karnatak points to the contrast between the beverage business, where average employee tenure is around six to seven years, and QSR, where the organisation can bring in 1,000–1,200 new employees in a month, with a similar number leaving.

Yet the group has pursued a common digital backbone since 2009, with a centralised IT function and a common ERP instance across its operations.

The philosophy is not to make every business identical. It is to give different businesses a common foundation on which they can operate and innovate.

That foundation has also helped RJ Corp scale. Karnatak recalls how the organisation moved from operating five or six plants primarily in northern India to expanding across other regions and other countries. He says the common IT backbone has enabled new plants and country operations to be brought onto the platform in roughly two to three months.

Technology, in this context, becomes less an IT function and more an architecture for organisational scale.

Putting intelligence in the hands of the frontline

One of the clearest examples is RJ Corp’s sales force.

Around 4,000 Pepsi market salespeople use Sales Force Automation tools that provide pricing, promotions, schemes and tax information directly to them. The system also generates suggested orders based on retailer profiles and past behaviour. Salespeople can photograph coolers to check compliance and use location data to confirm that customer visits are actually taking place.

The technology effectively removes routine cognitive friction. The salesperson does not have to remember every price or promotion. Instead, the system handles the information-heavy work, leaving the employee to focus more on the retailer relationship.

The same principle is visible inside QSR operations. Employee training is digitised, with staff required to complete modules and assessments before progressing. Kitchen orders contain the relevant recipes, reducing the scope for manual errors, while attendance is geo-enabled.

The broader ambition is clear: technology should augment the person closest to the transaction rather than simply automate the back office.

From enterprise data to enterprise intelligence

The underlying architecture is evolving as well. RJ Corp operates multiple systems spanning ERP, distribution management, SFA, demand planning, supply chain and enterprise performance management. Rather than expecting these systems to provide every answer individually, the organisation is bringing its data into data lakes and exposing that information through internal AI interfaces.

Currently, separate data lakes support sales and HR use cases, with additional applications being explored.

The group is also connecting manufacturing operations through IoT. Six or seven plants are already connected, with further expansion underway. Direct machine-generated data reduces the manual intervention and potential errors associated with entering information into ERP systems.

This is an important progression: ERP created digital records. Data lakes connected information. AI is now turning that information into an interactive decision layer.

AI is moving beyond the employee to the ecosystem

RJ Corp’s technology strategy also extends beyond its own workforce.

More than 6,000 beverage distributors can use digital systems to submit claims, view ledgers, and place orders. Its order management system operates around the clock and allocates inventory according to predefined logic rather than individual preference, reducing the scope for personal bias in allocation decisions.

Vendor invoice processing has also been automated. Invoices can be scanned and uploaded, interpreted by AI, matched against defined rules, and routed into the ERP for processing. Transporter selection uses reverse auctions, while a transport management system is being implemented to provide greater visibility into shipments.

Perhaps the clearest illustration is the proof-of-delivery process. A workflow that previously could take around 20 days to reach payment can, once the digital POD is uploaded and validated, potentially be processed within a day.

The result is an increasingly connected enterprise in which technology removes friction not just for employees, but for distributors, vendors and logistics partners.

Agents with a human hand on the switch

Karnatak’s approach to agentic AI is deliberately measured.

RJ Corp is already using AI to support recruitment, particularly for frontline roles. AI can match CVs against job descriptions, generate questionnaires, conduct adaptive interviews, and use multiple AI personas to assess candidates before an AI “judge” provides a final evaluation. The model is deliberately not being positioned as a replacement for human judgement in senior hiring.

The same principle applies to financial workflows. An AI agent can match an invoice against the purchase order and goods-received note and move it forward for processing. But final payment still requires human approval.

“We always want to put humans in the loop as far as critical decision-making.” For Karnatak, agentic AI is therefore not about handing over the enterprise to autonomous machines. It is about letting AI perform the analysis and groundwork while reserving consequential decisions for people.

The metric that matters

RJ Corp does not publicly quantify the financial benefit of individual AI initiatives. Instead, Karnatak says the organisation evaluates value across distributors, retailers, employees and management.

One measure stands out: revenue per employee. The philosophy is not to remove people simply because technology can perform some of their work. As the business expands, people can be redeployed towards higher-value activities.

If the business becomes significantly larger without a proportional increase in headcount, the organisation has increased its capacity.

“Revenue per employee is increasing; it means we are adding value.” This is perhaps the most consequential idea in RJ Corp’s AI strategy. The objective is not workforce reduction. It is organisational leverage.

The six-month CIO

Karnatak says RJ Corp no longer works with a traditional three-year technology roadmap. Instead, its planning horizon has compressed to six months to one year, reflecting the speed at which AI and other technologies are evolving.

But his bigger concern is cultural. Employees and consumers are already using AI outside the enterprise, changing their expectations of how services should work. RJ Corp is responding through education, responsible-AI awareness and its internal DISHA interface, which gives employees access to AI capabilities while keeping organisational data within its environment.

And Karnatak sees an even bigger opportunity: AI is becoming a new organisational resource.

Capabilities that once required an external partner—and months of understanding, development and negotiation—can increasingly be prototyped internally.

His philosophy is blunt: “Within the next 10 days, 15 days, let’s do it, show it, and then implement it.” That may ultimately be the real competitive advantage of AI for RJ Corp.

Not simply doing the same things faster. But turning ideas into working capabilities faster, putting intelligence in the hands of more people, and allowing a diversified enterprise to make better decisions at the speed of its customers.