Most admins say they need a major overhaul of data in order to succeed with AI (www.techradar.com)

Salesforce’s new survey paints a stark picture: 84% of business leaders say their data strategy needs a complete overhaul to succeed with agentic AI, and 76% feel pressure to deliver business value from data now. Respondents report pervasive quality and trust issues—about 26% of org data is considered untrustworthy, 42% lack confidence in AI outputs, and nearly half have drawn incorrect conclusions from poor context. Crucially, 70% believe their most valuable insights are trapped in roughly 19% of data that’s siloed or unusable, creating compliance risks, fractured customer views, and stalled decision-making. Salesforce’s executives frame this as a make-or-break moment: AI promise depends on stronger data foundations. Salesforce is pushing “zero-copy” architecture—letting AI access data where it resides without moving or duplicating it—as a practical remedy; 56% report early adoption and improved customer service and faster AI rollouts. For AI/ML teams the implications are concrete: invest in federated access, data virtualization, robust metadata, lineage and cataloging, continuous quality checks, secure access controls, and observability so models can reliably consume live, governed sources. Treating data and AI as an integrated strategy (data mesh/fabric, APIs, real-time pipelines) is now essential to move beyond pilots and scale agentic AI with trust, compliance and measurable ROI.