Most enterprise AI deployments fail not because of bad models, but because agents lack context. Hyland CEO Jitesh S. Ghai, CPO Michael Campbell, and Erie Insurance CIO Partha Srinivasa argue that unstructured data, the documents, images, and records companies already hold, is the missing layer between AI pilots and real business outcomes.
The conversation gets specific about where token allotment strategies break down and why feeding agents structured data alone produces shallow results. Erie Insurance serves as the case study, showing how a major insurer operationalized unstructured content across claims, underwriting, and customer workflows without the deployment dead ends that have plagued peers in banking and healthcare.
The argument here is architectural, not aspirational. Read the full transcript to understand how Hyland frames content management as infrastructure for agentic AI, and whether their approach to making unstructured data queryable at scale holds up under Kantrowitz's questioning.
[WATCH ON YOUTUBE →]