The IT Foundation AI Agents Actually Need: Data, Governance, and Access

AI agents are only as capable as the data and infrastructure beneath them. When data lives in silos, access is inconsistent, and no one has audited what AI vendors can retain or reuse, building agents simply automates the mess. This session walks through how Peninsula School District made data infrastructure the prerequisite: a central data warehouse with an MCP server layer enforcing role-based access control, a governed repository for unstructured data, and an isolated space where agents create artifacts without touching production systems. The district also renegotiated data-sharing terms with vendors and moved inference to AWS infrastructure with zero data retention, closing a privacy gap many districts have not yet identified. You will learn the specific questions to ask AI vendors and partners about data access and retention before signing anything. You will leave with a readiness checklist built on CoSN and ERDI frameworks to surface inconsistent data, unmanaged access, and silent vendor agreements so you can close those gaps before agents arrive.