Future-Proofing Your KM Initiatives
Agentic AI has had a remarkable effect on how KM functions within organizations. Future-proofing KM initiatives must take into account the many new possibilities and issues that only now are becoming apparent. As Sriram Iyer, senior VP and GM of AI strategy and products at OneTrust, explained, protecting future KM initiatives requires several key steps. “The most important step is to acknowledge and treat the knowledgebase as a strategic asset,” he asserted. “Knowledgebases were often viewed as important initially, but over a period of time came to be taken for granted, having achieved the goal of providing users with the information they needed. With the advent of AI and its associated transformation of businesses, it’s been necessary to rethink the role of KM as part of a dynamic system.”
Governance of content and processes will become both more important and more difficult. “The volume of information will continue to grow exponentially,” continued Iyer, “but the biggest difference will be that AI and AI agents will create and consume the majority of the knowledge.” If the data quality is poor and the context is not clear, AI agents will not be able to give the right answers. “The key will be to make information discoverable and trustworthy enough for the AI workflow, such that both humans and AI agents get the right answers,” he added.
The OneTrust AI-Ready Governance Platform is a unified system that modernizes enterprise oversight by connecting privacy, risk, data, and AI governance into a continuous, automated operating model. Workflows accelerated by AI provide uniform risk assessments across different data repositories, and guardrails such as data masking to control visibility to humans, AI, and AI agents having differing permission levels are enforced at the data layer. “Increasing concern about privacy, identity, and security is posing new challenges,” Iyer said. “It’s no longer just about checking a box, but about keeping up with a constantly changing environment.”
The precision required by AI in terms of accurate and retrievable information has given a boost to the issue of data quality. “AI will only be as effective as the information underneath it,” commented Iyer. “Significant weight has been given to the need for data security, and rightly so, but it is clear that data quality needs to be given equal importance in order for AI-driven KM initiatives to succeed.” In the past, questionable content might have been detected by a knowledgeable end user or subject matter expert. At today’s scale and speed of moving information, however, humans cannot compensate for poor quality data being produced or consumed by AI systems.
An additional point relates to legacy data. “This information should not be ignored even though it may use legacy APIs and not be MCP [Model Context Protocol]-compatible,” Iyer observed. “It needs to be brought along—including all its repositories and data schemas, whatever shape they are in— because the data chain is only as strong as its weakest link.” It is not enough to assume that every SharePoint repository has been blessed and tested, Iyer noted. “AI can access every nook and cranny, so it’s essential that audit processes be revamped to verify all the data.”