Overview
The enterprise landscape is buzzing with AI agents, rapidly deploying as copilots, assistants, and autonomous task-runners. McKinsey’s latest AI report reveals a surge in adoption: nearly two-thirds of companies were experimenting with AI agents by late 2025, with 88% using AI in at least one business function—a significant jump from 78% in 2024. Yet, beneath this enthusiastic embrace lies a critical challenge: only one in ten companies manages to scale their AI agent initiatives beyond early pilots. The core issue isn’t the sophistication of the AI models themselves, but rather the foundational data infrastructure supporting them.
Experts like Irfan Khan, president and chief product officer of SAP Data & Analytics, emphasize that delays in AI implementation stem from a lack of data architectures designed to deliver reliable business context to both humans and AI agents. In a rapidly evolving AI world, Khan stresses the urgency: “The only prediction anybody can reliably make is that we don’t know what’s going to happen in the years, months — or even weeks — ahead with AI.” To secure immediate victories, businesses must cultivate an ‘AI mindset’ and, crucially, ground their AI models with trustworthy, context-rich data. The next few months, or at most a few years, will be decisive in establishing this critical foundation.
Impact on the AI Landscape
This shift in focus—from model superiority to data architecture—profoundly impacts the AI landscape. While data has always been vital, its role in the age of AI agents is elevated to unprecedented importance. The true capabilities and scalability of agentic AI will increasingly be defined by the soundness of an enterprise’s data architecture and governance, rather than solely by the continuous evolution of the models themselves. To truly scale this transformative technology, businesses must adopt a modern data infrastructure that delivers not just raw data, but data imbued with meaningful business context.
Traditional data views often categorize structured data as inherently high-value and unstructured data as less valuable. However, AI agents complicate this distinction. For agents, high-value data is less about its format and more about the depth of its business context. Data critical for functions like supply-chain operations or financial planning gains its true value from context. Even fine-grained, high-volume data from IoT devices, logs, and telemetry, while seemingly rich, only yields significant value when delivered with relevant business context. As Khan points out, the real risk for agentic AI isn’t a scarcity of data, but a lack of proper grounding. This deficit contributes to what the Institute for Data and Enterprise AI (IDEA) terms ‘trust debt,’ with two-thirds of business leaders not fully trusting their data.
Practical Application
For businesses looking to operationalize AI agents effectively, the path forward is clear: prioritize context over mere data volume. “Anything that is business contextual will, by definition, give you greater value and greater levels of reliability of the business outcome,” Khan asserts. This means moving beyond simplistic classifications of data value. Both structured transactional data and repetitive unstructured data can hold immense value when properly contextualized for AI agents.
Deriving this essential context involves several practical approaches: seamless integration with existing software systems, on-site analysis and enrichment processes, and robust governance pipelines. These methods transform raw data into intelligent, actionable insights that AI agents can reliably interpret and utilize. Data that lacks these qualities will inevitably be untrusted, hindering agent performance and business outcomes. Addressing this ‘trust debt’ is paramount. By investing in modern data infrastructure that emphasizes context, businesses can ensure their AI agents are not just deployed, but are also effective, reliable, and scalable, unlocking the full potential of their AI investments. The time to build this intelligent data foundation is now.
Original source: View original article
