AI Success Depends on Trusted Real-Time Data, Says Confluent's Rubal Sahni

Artificial intelligence has moved well beyond experimentation. Enterprises have invested heavily in large language models, copilots, and agentic AI, but many continue to struggle when it comes to deploying these technologies at scale. Confluent's 2026 Data Streaming Report, based on responses from 4,625 IT leaders across 14 countries, including 650 from India, suggests the bottleneck is no longer AI investment. Instead, organisations are finding that fragmented data, legacy architectures, and the lack of real-time data infrastructure are slowing AI adoption. The study found that 79% of Indian IT leaders believe inadequate real-time data infrastructure is delaying AI growth, while 78% cite insufficient infrastructure for real-time data processing as a major challenge. Only 37% of organisations have successfully deployed agentic AI into production. In an interaction with CIOL, Rubal Sahni, AVP – India and Emerging Markets, Confluent , discusses why AI success increasingly depends on trusted, real-time enterprise data; how organisations are measuring returns from data streaming; and why future AI leaders will be distinguished less by the sophistication of their models and more by the strength of their data foundations. Interview Excerpts The report suggests that AI infrastructure, rather than AI investment itself, has become the biggest bottleneck. Have enterprises collectively underestimated the importance of real-time data infrastructure while overestimating what AI models alone can achieve? Where do you see this gap most clearly? Over the last couple of years, most organisations have invested heavily in AI experimentation. What many are now discovering is that scaling AI requires a very different set of capabilities than building a successful proof of concept. Our research shows that 79% of Indian IT leaders believe a lack of real-time data infrastructure is slowing AI growth, while 78% cite insufficient infrastructure for real-time data processing as a key challenge. These findings reinforce that AI success depends on continuously supplying models with fresh, trusted, and contextual enterprise data. We see this most clearly when organisations move into production. AI systems are expected to respond to live customer interactions, operational events, and changing business conditions. If the underlying data arrives late, lacks context, or remains locked across disconnected systems, AI cannot make reliable decisions. At that point, scaling AI becomes as much a data challenge as an AI challenge. Many organisations have successfully experimented with AI but continue to struggle with production-scale deployments. Beyond real-time data, what are the biggest roadblocks today—legacy architecture, data governance, fragmented ownership, skills, or organisational culture? Many organisations have already demonstrated that AI works in controlled environments. Scaling those successes across the enterprise is where the real challenge begins. Our research found that 79% of Indian IT leaders encountered at least three challenges while scaling AI initiatives, suggesting this is rarely a single-point issue. Alongside insufficient real-time infrastructure, organisations struggle with uncertainty around data lineage, timeliness, and quality, as well as fragmented ownership of enterprise data. Legacy architectures add another layer of complexity because many enterprises still rely on batch-orientated systems designed for periodic reporting instead of continuous intelligence. Organisations making the fastest progress are addressing infrastructure, governance, and organisational alignment together. That combination creates the trusted data foundation required to operationalise AI at enterprise scale. Nearly 87% of respondents reported a two-to-five times return on investment from data streaming. How are enterprises measuring that value? Is the conversation moving beyond cost savings toward business agility, customer experience, and faster decision-making? Organisations are increasingly evaluating data streaming across a much broader set of business outcomes. Infrastructure efficiency and modernisation remain important, but enterprises are now measuring improvements in customer experience, application resilience, fraud detection, operational efficiency, and decision-making speed. For customer-facing businesses, resilience is becoming one of the clearest indicators of value. If a revenue-generating application goes offline, organisations measure not only infrastructure costs but also business disruption, customer trust, and recovery time. Real-time data enables teams to detect problems earlier and respond much faster. Financial institutions, meanwhile, often evaluate returns through faster fraud detection and stronger real-time risk management. Your report highlights data quality and infrastructure as major barriers to agentic AI. As autonomous AI systems become more common, how should enterprises rethink governance, accountability, and trust when AI starts making decisions rather than simply generating insights? As organisations deploy agentic AI, governance becomes significantly more important because these systems are increasingly expected to take actions rather than simply generate recommendations. Our research shows 72% of IT leaders identify data infrastructure and data quality as barriers to agentic AI adoption, while only 37% have agentic AI running in production today. Every decision an AI agent makes depends on the quality, timeliness, and context of the data it receives. Organisations, therefore, need confidence in where data originates, how current it is, who has access to it, and how it is governed throughout its lifecycle. They also need clear operational guardrails defining which systems AI agents can access. Well-designed governance frameworks improve both trust and operational efficiency. India appears to be at an inflection point for real-time data adoption. Is this shift being driven primarily by AI ambitions, cloud maturity, regulatory pressures, or changing customer expectations? Which sectors do you expect to lead this transition, and why? India is experiencing several technology shifts simultaneously. AI adoption is increasing demand for trusted enterprise data; cloud adoption is enabling modern architectures; regulatory expectations around governance continue to evolve; and customers increasingly expect real-time, personalised digital experiences. Our research reflects this convergence. Eighty-six percent of Indian IT leaders say enterprise data has become a top priority for AI, while 92% rank data streaming alongside AI as a strategic investment priority. We expect banking and financial services, retail, e-commerce, quick commerce, telecommunications, and other digital-native businesses to lead this transition because they already process millions of real-time events every day. Their ability to react instantly directly affects customer experience, operational efficiency, and business growth. Looking ahead, will enterprise AI leaders be defined by the sophistication of their AI models or by the strength of their real-time data architecture? What will separate organisations that successfully operationalise AI from those that remain stuck in pilot mode? AI is entering a phase where success will increasingly be determined by execution rather than experimentation. Organisations making the most progress begin with clearly defined business problems, validate AI through focused pilots, and then invest in the data foundation needed to scale those solutions across the enterprise. AI models will continue improving across the industry. The real differentiator will be an organisation's ability to connect those models to fresh, trusted, and contextual enterprise data. That is what enables faster decisions, reliable automation, and measurable business outcomes. There is growing discussion that every enterprise now needs an AI strategy. Based on what you're seeing across customers, would you argue that organisations actually need a real-time data strategy before they need an AI strategy? Why? AI strategy and real-time data strategy are increasingly becoming inseparable because they ultimately support the same business objective, creating value from enterprise data. Our research found that 97% of respondents believe data streaming increases the impact of AI investments, while 91% rank data streaming as a strategic priority alongside AI. Organisations making the fastest progress are not choosing one strategy over the other. They are developing AI and real-time data strategies together from the outset because each strengthens the other. AI determines where intelligence creates value, while real-time data ensures AI has the trusted, contextual information needed to operate reliably at scale.