Artificial Intelligence (AI) has become essential for businesses striving to maintain relevance in today’s fast-paced market. The strategic implementation of AI goes beyond mere technological enhancement; it is now a fundamental driver of business evolution. Organisations that harness AI’s capabilities effectively can unlock unprecedented potential, transforming customer engagement, operational efficiency, and innovative business models.
Furthermore, the significance of AI extends into the boardroom, where it demands attention not as a futuristic novelty but as a critical business imperative. The executive leaders, Ramesh Shanmuganathan and Rajeev Singh, emphasise that the true value of AI is realised through its integration into the core of business strategies. They discuss how focusing on data-centric approaches, rather than merely adopting AI, is crucial for business sustainability, especially for mid-sized and emerging firms in South Asia.
AI enables hyper-personalisation, meeting customer expectations by delivering tailored experiences based on individual behaviours and preferences. It also drives process efficiency, capturing and building upon knowledge to optimise operations continuously. Moreover, AI facilitates business model innovation, urging companies to rethink and reinvent their methods of value creation. However, the success of these endeavours hinges on the quality of data that AI systems utilise.
In this rapidly digitalising era, the development of AI parallels the human cognitive process, evolving and improving through data consumption and pattern recognition. As AI advances, it increasingly mirrors and even surpasses human cognitive capabilities, positioning itself as an indispensable asset in the modern business landscape.
The transformation driven by AI is evident in regions like India and the broader subcontinent, where digital adoption has spurred significant productivity gains. Yet, businesses that lag in embracing this digital framework risk missing out on growth opportunities and a competitive edge.
AI is not just a competitive advantage. It is a business imperative and should be a key item on the boardroom agenda. Ramesh Shanmuganathan, EVP/CIO of John Keells Holdings and CEO of John Keells IT, and Rajeev Singh, VP of Corporate Business for SAP India, discuss AI’s shift from a technological advantage to a boardroom necessity. They emphasize that data-centricity will be crucial for business viability, especially for mid-sized and emerging firms in South Asia. Their conversation covers moving AI from hype to execution, embedding it into business models, and preparing people and systems for digital transformation.
AI is often touted as a game-changer. However, what makes a successful AI strategy in today’s context? Why is data-centricity more critical than simply having an AI-first approach?
Ramesh: AI is one of the most overused terms today. It was seen as a competitive advantage a few years ago. However, today, it’s a business imperative. There are three key reasons for this:
Customer expectations: Customers now expect hyper-personalisation. They don’t want random products or content thrown at them. They want experiences tailored specifically to them. Think about platforms like Netflix or Amazon: based on your behaviour and purchase patterns, they personalise recommendations, which increases your likelihood of engaging or buying.
Process efficiency: AI is helping organisations drive process efficiency. Many inefficiencies exist because people take the knowledge with them when they move on. With AI, companies can capture that knowledge, build on it, and keep improving over time, making operations smoother and smarter.
Business model innovation: AI is pushing companies to rethink how they operate and create value in new ways. However, none of this is possible without quality data. AI is only as good as the data behind it.
How do personalisation, process efficiency, and business model innovation rely on data? Can you break down how data underpins each of these outcomes?
Ramesh: Like any human being, a baby is born with a blank mind. Over time, through experience and learning, the brain develops, and memory becomes richer. AI works similarly. The more data it consumes, the more it learns. It starts recognising patterns, understanding algorithms, and eventually begins to think on its own, especially with advanced forms of artificial intelligence. That’s the direction the world is heading in: we’re trying to replicate the human brain using computing. Over time, AI may outperform the average human brain in many areas. We’ve already started to cross that threshold. That’s the parallel I’d draw: AI is evolving just like a human but much faster.
India and the broader subcontinent have seen impressive productivity gains driven by rapid digital adoption. What are the risks and missed opportunities for businesses that have yet to embrace this digital architecture, where technology is central to driving productivity and growth?
Rajeev: Adopting AI is no longer optional for traditional businesses—it’s a necessity. Just look at the Fortune 500 from 2000: over half of those companies no longer exist, while those adapted to change have thrived. Today, AI is that change. But ambition alone isn’t enough. Many companies jump on AI trends without a clear strategy, which can be risky. Brick-and-mortar businesses should start by identifying a specific problem to solve and then build an AI strategy around it. Without that foundation, efforts will likely fail. AI brings huge potential—but also responsibility. Companies must upskill their teams, protect their competitive edge, and be mindful of how they use and share data, especially with third-party AI platforms.
If you’re speaking to a mid-sized company pulling in over $10 million annually and eyeing $100 million, what’s your pitch? Ambition alone won’t get them there—where should leadership start, and what core fundamentals must be in place to scale sustainably and transform effectively?
Rajeev: Companies should start small and stay focused. The first step is identifying clear, practical problems to solve. The ultimate goal should be to build a strong foundation, the application layer, where core business processes like finance, supply chain, and procurement are fully digitized and generate structured, coherent data. Applying AI at a business level without this foundation will fail to deliver real results. Tools like OpenAI may help with emails or images, but reliable business outcomes—like cash forecasting or supplier risk—depend on solid systems underneath. This is especially relevant in the SME segment, which comprises 80% of our global customer base. In the Indian subcontinent, SMEs are typically defined as companies with revenue under $250 million—and we work with many in the $25 million to $1 billion range.
Ramesh: If you think about building a business today, it’s like building with Lego blocks—you don’t have to create everything from scratch. This model works exceptionally well for companies starting with $25 million in revenue, who might only focus on one part of the broader value ecosystem. By being digital and leveraging AI, these companies can optimize precisely what they need from that ecosystem. Take the example of COVID-19—suddenly, every kitchen could become a cloud kitchen. However, that shift wouldn’t have been possible if platforms like PickMe or Uber hadn’t already handled last-mile delivery. These businesses didn’t need to own the entire delivery chain but plugged into the ecosystem using digital tools and AI to get the desired outcomes. That’s the beauty of today’s business model innovation: you no longer have to own the entire value chain. Instead, you can focus on the one or two things you’re good at and use digital platforms and AI to connect the dots and optimize the rest.
Bringing it back to data-centricity, how should SMEs think about scaling in relation to application readiness and data strategy? Are these distinct priorities, or do they need to be addressed in tandem to lay the groundwork for AI, automation, and sustainable growth?
Ramesh: Take a manufacturing example, like a soft drink producer managing its supply chain. A common practice might be to purchase raw materials like sugar in bulk, perhaps every three or six months, based on what they consider efficient sourcing. That might seem smart, mainly if they’re used to planning around global market prices. However, they might not realize that this approach could limit their flexibility and cost savings.
Imagine they had access to a procurement system like SAP Ariba, which gives visibility into a global network of sugar suppliers. With that real-time insight, they could make more informed decisions, maybe even source sugar just in time at better prices rather than locking in bulk purchases every few months. Relying on traditional practices might feel comfortable, but without tracking price volatility or supplier dynamics through a system like Ariba, they’re potentially missing out on significant savings and agility in procurement.
You’re sharing data across the network to optimize your decisions. Can you explain how this happens?
Rajeev: Building on data centricity, AI is only as good as the data it’s trained on. Public models like OpenAI are trained on general data from Google, X, and the news, but that’s not always relevant for specific industries. For example, pharma has complex compliance and sourcing needs that generic AI can’t grasp. That’s where industry-specific AI becomes critical. For AI to be truly useful in business, it must be relevant (understanding your industry), reliable (trained on quality, specific data), and responsible (traceable and auditable). All this depends on having a strong data foundation from your core systems. SMEs often lack the volume of data to train robust AI models themselves. That’s where platforms like SAP help. Its AI is trained on data from over 40,000 enterprises across industries. So, when an SME adopts it, the insights are already contextual, relevant, and immediately actionable.
Ramesh: For those who have worked with large language models (LLMs), we’ve probably noticed that the responses to very specific questions tend to be broad and generic. However, when you benefit from working with data from 40,000 customers, as SAP does, the insights are far more tailored and actionable.
Larger firms often treat data as their ‘new oil’, a proprietary asset that fuels their competitive edge. But in today’s digital landscape, even a $20 million company in India can tap into similar insights using available tools and platforms. How do you address the concern that this levels the playing field, and what does that mean for both large and small companies thinking about data strategy going forward?
Rajeev: First, it’s essential to understand that the data is precise to each customer. The models are trained on patterns from that particular data, not on data from all 40,000 customers. For example, cash position forecasting in a pharmaceutical company requires different inputs compared to banking or manufacturing. The AI model understands that specific factors need to be considered for each industry. So, when an SME uses this AI, the model knows exactly what pattern to look for based on their industry. Crucially, the SME’s data remains private and stays with them. They’re leveraging the AI model or algorithm, not sharing their proprietary data.
Ramesh: There’s a common misconception that software like SAP or DeepSeek compromises your data, but that’s not true. This can be a concern with generic AI services, especially those of the cloud. However, when using your instance of SAP, the AI algorithms run directly on your data, keeping it private and secure. The model is trained on global datasets, allowing you to benefit from insights and validation from a wider pool of knowledge without exposing your data, which remains confidential and not part of the public domain.
What do local businesses gain by partnering with a company like SAP? And when working with a partner like JKIT alongside SAP, what kind of synergy do you bring to the table?
Ramesh: John Keells was one of the early adopters of SAP in 2004 when it was seen as a solution only for large enterprises. But that perception has changed. Today, SMEs embrace SAP because it’s more than just transaction processing. It offers flexibility, modularity, and powerful tools. For example, manufacturers can use external procurement portals and integrate seamlessly with SAP. Its Lego-like architecture lets businesses scale gradually. Previously, SAP required heavy infrastructure like data centres, servers, and storage. Now, it’s all cloud-based. You subscribe and access the service, with SAP handling R&D and updates. It’s like buying a beer. You don’t need to build a brewery. This model enables even $25 million-revenue companies to benefit from SAP’s efficiency, innovation, and scalability without the traditional cost burden.
JKIT has long been an SAP implementing partner. What are your most significant challenges in implementing AI solutions, and how have you overcome them?
Ramesh: Cloud adoption is the biggest challenge, especially when comparing Sri Lanka to more mature markets like the USA. In Sri Lanka, we’re at less than 25%, whereas the US is close to 90%. The issue is that most of your infrastructure is running on-premises in your data centres; you’re constrained by capacity and a capital life cycle. You have to account for asset depreciation every five years, and if you need to increase capacity, you face a wait cycle of three to six months for order fulfilment. This significantly limits innovation.
That’s why the first step we need to take is moving to the cloud. Today, SAP no longer offers on-premises solutions; everything is cloud-based. If a business starts today, there’s no option but to go to the cloud. The real challenge lies in transitioning legacy systems to the digital world. Many companies still fear the unknown and hesitate to let go of their legacy tech to move to the cloud. However, if they don’t take that first step, leveraging AI will be daunting and challenging, as their data won’t be ready to fuel innovation.
The second challenge is data quality. Do you have the correct data to run AI effectively? Many companies deploy AI without ensuring they have clean, accurate data, and poor data quality becomes a major roadblock. So, the two biggest challenges are cloud adoption and data quality.
Cloud adoption is 10% in Sri Lanka. Does India face a similar challenge, and how are you approaching that?
Rajeev: We recently surveyed 2,500 enterprises, and the results showed that 96% of SMEs consider AI a top priority. As for the difficulty of implementing an AI strategy, especially for SMEs, large enterprises in India are more likely to leap because they have sufficient data and established processes. However, small and medium enterprises are still refining their operations and data to leverage AI. One major challenge is that while everyone wants to adopt AI, many businesses lack a clear use case. AI adoption becomes a significant hurdle for any AI strategy without a solid business case. I’d also like to highlight the growth of SAP in Sri Lanka. When we entered the market in 1998-2000, it’s been 25-26 years, we wouldn’t have been able to grow to over 300 customers in Sri Lanka without local partners like JKIT. This has been SAP’s strategy globally. While we provide the product, we rely on regional partners to take the message to large and small enterprises in each country.
In the next 5 to 10 years, what trends will shape AI’s future in business?
Ramesh: The most significant shift I see is AI moving to the edge – Agentic AI. Today, much of AI runs in large data centres like hyperscalers. However, with the growing computing capacity, AI at the edge is poised to be a game-changer. It will bring significant productivity benefits and provide instantaneous service to customers autonomously. For example, consider an ATM. It can predict its refilling needs based on the cash it dispenses. It doesn’t need a person at the bank to analyze that data; the ATM can track daily transactions—like how many 50 or 100-rupee notes it dispenses—and automatically adjust for seasonality. Similarly, a point-of-sale system could predict the required stock replenishment in a supermarket. This shift will drive innovation and improve efficiencies, particularly in supply chains. With AI at the edge, you can optimize stock holding and inventory costs, reducing waste and enhancing overall operational efficiency. The power of AI lies in using data at the edge rather than sending everything back to the core for processing.
Can you define what the concept of the edge strategy is?
Ramesh: When I refer to the edge, I mean where your product or service is consumed. For example, an ATM is used by a bank’s customers, a point-of-sale system in a supermarket is for retail consumers, and manufacturing could be a store. Take Sysco, for instance. They run their largest offshore development centre in Sri Lanka and are also one of the most prominent players in the distribution of food supply chains. Their first deployment involved point-of-sale machines. These machines provided data about what restaurants were ordering on any given day. Sysco Labs, which started around 10-15 years ago, used that data at the edge to fuel innovation in managing the entire food supply chain.
Businesses need to consider why they’re collecting data and how that data impacts their end consumer at the edge. If you don’t connect these dots, AI won’t be sustainable. That’s why point solutions for AI tend to have short lifecycles, requiring constant reinvention to stay relevant. Ultimately, the goal is to personalize the customer experience so they remain satisfied and keep coming back.
So, do these AI insights necessarily lead to an altered strategy?
Rajeev: Another term gaining traction in tech is agentic AI. This concept relates closely to edge computing. For instance, imagine you’re in an amusement park, and an AI is monitoring the health of your rides. If the machine sends data to the cloud, there’s a delay as the cloud analyzes the data and warns that the machine is about to fail. That’s latency. Ramesh refers to an edge solution, where a chip directly on the machine can instantly detect an issue and stop it before failure occurs. This is where the future of AI is heading. As a tech company, we focus on solving “agentic AI at the edge.” These agents communicate with the backend for decision-making and handling information and transactions. In SAP, for example, we have already applied this concept. We have agents across the system: an agent for the supply chain, an agent for customer service, and an agent at the point of sale. These agents continuously check inventory levels, pinging each other to confirm stock availability. If something’s wrong, the transaction happens automatically: inventory is dispatched, and the issue is resolved.
We see this as the next level of AI: AI that functions in a co-pilot mode, helping you act quickly without inputting every decision manually. For example, SAP’s Jewel is a co-pilot assistant. It doesn’t replace you, but it guides and assists, asking questions and helping you make informed decisions without constant oversight.
Ramesh: That’s why we call it a co-pilot: the pilot is still there, but the co-pilot assists you. It’s an assisted experience. AI also has a lot of opportunities to take things to the next level. Take, for example, cybersecurity. Today, endpoint security on your devices draws intelligence from a central node and can even mitigate threats. This is where we’re heading with MXDR (Managed Extended Detection and Response) and defence remediation. Technology is evolving in this direction, where the central brain still exists, but the edge can make decisions with the limited knowledge and algorithms it’s been programmed with.
How can medium-sized organizations in Sri Lanka, particularly SMEs, prepare for the potential disruptions that AI might bring to traditional business models, especially considering that many are currently unprepared for such technological shifts?
Ramesh: The more significant challenge will be accepting and getting comfortable with the evolving technology rather than resisting it. The more you alienate technology, the more you risk threatening your survival. The key is to start familiarizing yourself with and experimenting with the technology. Begin with low-hanging fruit to gain experience and understand that foundational elements, such as a solid transaction system and quality data, are required even in using AI. Building this foundation will likely take three to five years. However, the good news is that platforms like SAP already have AI embedded today. So, when you begin using such platforms, your AI journey starts immediately, without waiting for the full three to five-year development.
Rajeev: The fastest way to adopt AI is by using a platform with AI already embedded. Building data lakes and manually connecting data is complex and often strips data of its context, like knowing a material number but not its source or purpose. Rebuilding that context takes time. With platforms like SAP’s ARP, data is already contextualized. It understands key relationships like a phone linked to an employee or supplier, which helps AI learn faster. This significantly speeds up adoption. For SMEs, using an ERP system like SAP’s Grow, which comes with embedded AI, is a smart move. It covers core processes, and you can integrate non-SAP data if needed, allowing AI to learn across both environments and accelerate transformation.
Ramesh: They have 150 pre-built models, like mini LLMs, tailored for roles such as CFO, CIO, and CHRO. For example, a CFO dashboard displays key performance indicators (KPIs), and the CFO can easily drill down into the data. As I mentioned earlier, AI is about asking the right questions, not just finding answers, and all those questions are already embedded in the system. This is known as embedded analytics. Building something like this from scratch would require data scientists, modellers, and other resources, which are typically unaffordable and unsustainable for SMEs. This solution provides a quick path to evolution, and once you’ve gained some traction, you can collaborate with partners to build more purpose-built models as extensions.
What capabilities and skills should an organization consider when starting its AI journey? Does the human factor play a critical role in this transformative journey?
Rajeev: Nothing can be implemented without the right people skills. People need to embrace this change. From a skills perspective, roles that used to focus purely on coding will need to upskill because AI can now make coding 10 times faster. From a people strategy standpoint, employees must leverage these tools to become more productive. If they’re not ready for this shift, all implementations will likely fail. Upskilling and reskilling are essential. However, while it’s possible to do this in-house, the first step should involve working with a reliable partner, like how we collaborate with John Keells. Once you have that support, you can train and continuously upskill your internal team. People development is key to success in this journey.
Ramesh: Take a simple example: 10-15 years ago, if you were planning to go somewhere, people would debate the best route to take, and whoever drove better might claim they’d get there faster, even if the path were longer. Today, with Google Maps, all the data sets allow us to determine the most efficient route based on traffic conditions and the time of day. This is how context-sensitive decisions are now. AI today provides information, but the final decision still lies with the human, the ‘pilot.’ If I have access to Google Maps but choose not to use it, that’s my choice. The key is that AI helps optimize decisions, outcomes, and costs, leading to more efficient results. People need to understand that AI is not necessarily superior, but it offers better data and algorithms to improve decision-making. It’s about getting used to the technology for the right reasons, not the wrong ones. Ethical guardrails and the right mindset are crucial in this journey.
As Sri Lanka embarks on a nationwide effort to digitize identity, what should businesses consider regarding the potential implications for their industries? With more people preferring mobile smartphones for payments, should companies consider incorporating AI into their solutions to adapt to these changes?
Ramesh: Definitely. If you look at the healthcare, agriculture, and tourism industries, AI could significantly eliminate friction points and help accelerate growth. In sectors like agritech and healthcare, where there is often a shortage of people but high demand, AI can be a quick fix through process automation or enabling self-service. The key is to follow the ‘mobile first, cloud first, AI-first’ approach. Without mobile, there’s no point in moving to the cloud; without the cloud, you can’t effectively leverage AI. We advocate for linking mobile, cloud, and AI because, when integrated, businesses can offer consumers the flexibility to access products or services wherever and whenever they want. With AI, companies can predict customer behaviour, such as which store they’ll visit, what day, time, and quantity.
While it may seem overwhelming for businesses that have yet to start, does this mean they have to wait five years to gradually implement a ‘mobile first, cloud first’ strategy, or is there a way to accelerate their journey with pre-packaged solutions?
Rajeev: We’ve implemented our fastest cloud AI-based ERP solution in 12 weeks with many small and medium enterprise customers. This is the pace you can expect with an SAP solution. It’s typically a 3-4 month process to be fully operational for SMEs. On the topic of a digitalized economy and unlocking potential, we’ve seen how initiatives like ONDC and UPI in India have transformed the digital landscape. These efforts have created enormous opportunities for SMEs. However, if SMEs aren’t prepared, they could easily be overwhelmed by the rapid influx of digital demand. When the digital economy opens up, there will be overwhelming data, supplier requests, and customer inquiries. This is why having a solid AI-driven platform strategy is critical. Without it, businesses may waste valuable resources chasing the wrong leads or suppliers. The key is to be ready to effectively navigate and capitalize on the massive opportunities that digital economies will unlock.
Ramesh: Singapore and Dubai are particularly important examples. To truly drive digital transformation, you need to focus on e-government. With AI, you can quickly automate many processes, eliminate inefficiencies, and reduce bureaucracy, ultimately empowering consumers to access government services more effectively.
Is there anything you would like to add in conclusion?
Rajeev: Any organization pursuing digital transformation or AI should focus on four key areas. First, define your purpose: Clearly understand why you’re transforming, whether it’s to grow revenue, enter new markets, or improve operations, and align your KPIs accordingly. Second, build a strong foundation: A solid ERP or application layer is essential to support your transformation. Third, ensure quality data: Accurate, well-managed data is critical for meaningful AI insights. Avoid bad practices like backdating entries that can compromise data integrity. And fourth, upskill your team: Technology and business are now deeply connected. If your team isn’t upskilling and adapting, the transformation will fail.
Also, leverage your ecosystem by learning from peers and partners who’ve done it before. Don’t go it alone. These are the fundamentals for a successful digital journey.
Ramesh: I want to revisit my opening statement: Today, AI is no longer a competitive advantage. It is a business imperative and should be a key item on the boardroom agenda. Too often, it’s left to the IT or tech teams to figure out what to do with AI, but this should not be the case. AI is a business issue that every board member and C-level executive should be focused on.
This shift will raise essential questions: Is the organisation mature enough to embrace AI? Do we have the right people to steer it? Do we have the right use cases to apply them on? Do we have the right technology to drive AI? Do we have the correct data to fuel it? These questions need to be asked at the board level, challenging the organisation to ensure it’s not kicking this agenda down the road or falling behind in terms of industry adaptation.
AI, like e-commerce, cloud computing, and security in the past, should be driven from the top down. Private or public organisations must be concerned about doing it right. Just as security and AI are now intertwined and serious boardroom topics, they should no longer be avoided in discussions.