Dark Data: Agentic AI's Biggest Bottleneck Revealed

AI systems in large organizations can only access 45 percent of company data. This figure drops to 33 percent in financial services and 40 percent in healthcare. These findings come from a new report based on a global survey of 300 senior executives. Only 51 percent of these executives trust the accuracy of their AI agents' decisions. In organizations where AI accesses over 70 percent of data, all executives report mostly accurate decisions. However, among "data laggards," where AI reaches 30 percent or less, only 22 percent trust these decisions. Many organizations face challenges with legacy data systems, which prevent scaling agentic AI and cause delays. Over half of all respondents say their current data platform is the main bottleneck. The report identifies this inaccessible information as "dark data," which includes unstructured content like images and video, as well as structured data in operational systems. Entrenched data silos are the biggest platform challenge, cited by 50 percent of respondents. Despite these issues, 83 percent of organizations already use AI agents to some degree, and all surveyed organizations expect to use them within two years. This shows the critical need to address data access for effective AI deployment.