AI models are becoming increasingly accessible, but turning them into products that work reliably remains a challenge, particularly when enterprises are dealing with fragmented data infrastructure. At TechSparks 2026, Michael Cronin, Vice President, Sales, Asia-Pacific, Japan & India, Couchbase, joined Shivani Muthanna, Senior Director – Strategic Content & Video, YourStory, for a discussion on moving AI beyond experimentation, production failures, governance challenges, customer experience, and India's growing role in AI innovation. AI’s missing link: the data layer Over the decades, large enterprises have adopted different databases and point solutions to address specific problems. While each solved an immediate need, the result has often been a fragmented data layer, with multiple systems that struggle to work together. Cronin sees this fragmentation as a major barrier to AI adoption. Citing industry estimates, he said 70–80% of AI development projects never reach production, while an MIT study found that 95% fail to become profitable. Couchbase’s operational data platform is designed to address this challenge by helping enterprises consolidate their data infrastructure. Rather than trying to fix everything at once, Cronin advised companies to start small. “My advice to founders and enterprises that are struggling with the fragmentation of the data layer is to pick your most critical use cases first,” he said. Cronin cited the example of Tesco, the UK retailer with a large global Capability Center (GCC) in Bengaluru. Its multiple systems were affecting customer experience, particularly during busy shopping periods such as Black Friday and weekends. Tesco began consolidating its data layer, starting with a single instance and use case before extending the effort across its technology stack. The consolidation enabled the retailer to handle millions of transactions per second with sub-millisecond latency. According to Cronin, the example shows how companies can begin addressing fragmentation through a single use case before expanding the changes across their operations. AI moves at machine speed. Can data keep up? The pressure to develop AI products has led many enterprises to prioritize models and applications. However, pilots often run on small datasets in controlled environments, leaving organizations unprepared for the demands of production. Cronin pointed to the growing scale of machine-generated traffic. “Eighteen months ahead of schedule, machine traffic has overtaken human traffic on the internet. That’s the AI scale. That’s the machine speed scale that enterprises need to operate on,” he said. When AI applications encounter real-world workloads, fragmented data systems can prevent them from operating reliably at scale. The growing use of AI agents adds to the challenge. As agents become more autonomous and take on decision-making responsibilities, enterprises need to ensure they can access the right data while working within appropriate governance frameworks. Fragmented systems can make communication and governance more difficult, limiting the ability of agents to operate autonomously. Cronin also urged enterprises to reconsider how they allocate their AI investments. While models are evolving rapidly, enterprise data remains a critical asset. “The models are rentable. Today’s best model is yesterday's newspaper. The true asset for the enterprise is the data,” he said. More API layers aren't the answer As enterprises began recognizing that fragmented data was holding back their AI initiatives, many responded by adding APIs and connectors to link existing systems. Cronin cautioned that this approach could introduce further complexity rather than solve the underlying problem. Each additional connection brings latency, risk, and another potential point of failure. Instead, he recommended consolidating the data layer so applications and AI agents can communicate with it more directly, bringing data closer to customers. According to Cronin, many enterprises, including large GCCs in India, are now reassessing their architecture to address data fragmentation. Without these changes, further investment in AI development could lead to the same problems. Consolidating data can also improve customer experience by helping systems manage growing demand, remain available and respond quickly. As customers expect more from AI-powered services, having data available at the edge is becoming increasingly important. India: From back office to innovation leader Cronin observed that business cultures vary across the Asia-Pacific region, but fragmented data infrastructure remains a common challenge for AI adoption. At the same time, India is becoming an important centre for AI innovation and decision-making. He noted that 80% of Couchbase's top 50 global accounts have a significant presence in India, with much of their AI innovation and leadership based in the country. Couchbase is also expanding its presence in India, with its Asia-Pacific headquarters moving to Bengaluru. Its Indian enterprise customers include Air India, Airtel, BigBasket, and Mahindra Finance. According to Cronin, India’s role in global business has moved beyond cost savings, with innovation, leadership, and important business decisions increasingly taking place here. He cited Air India's next-generation entertainment platform, which is being built on Couchbase. The platform aims to provide passengers with personalized shopping and entertainment experiences even when an aircraft is offline. For Cronin, the project shows how Indian enterprises are using technology to improve customer experiences under real-world operating conditions.