Oracle announced new capabilities for its AI Data Platform that give AI agents a shared understanding of business entities, relationships, policies and workflows, while adding tools to manage and govern an organization’s growing AI estate.

The capabilities were announced at Oracle AI World in Las Vegas on October 27. Oracle said they are expected to become available within the next 12 months.

The platform uses business ontologies to capture how organizations define data, relationships and policies. An AI Discovery agent can generate those ontologies by analyzing database schemas, semantic models, application and API documentation, and information in unstructured documents. Business domain experts review, refine and approve the resulting context.

The company also introduced Oracle AI Registry, a centralized inventory of AI agents, tools, skills and Model Context Protocol (MCP) servers. It records ownership, versions, dependencies, permissions and lifecycle status, giving organizations visibility into what AI capabilities exist and who controls them.

A planned AI Governance and Control capability will add testing, policy assessment and ongoing monitoring of agent behavior, quality, performance, usage and cost.

Agent Inventory and Access Controls

For chief data officers (CDOs), the announcement reflects a growing challenge as organizations move from individual AI applications toward large estates of interconnected agents.

Keeping track of which agents exist, what data they can access, who owns them and what permissions they have becomes increasingly important as agents begin taking actions across business systems.

The Oracle AI Registry is of particular importance, as it’s designed to provide that visibility. Business users can discover and work with registered agents through Oracle Agent Hub or supported third-party AI assistants, including ChatGPT and Claude, while access and execution remain subject to applicable enterprise security policies and controls.

This makes AI governance increasingly connected to data context and access management. An agent needs to understand what enterprise data ‘means’, but organizations also need to establish which data it can use, which actions it is authorized to perform and who remains accountable for those actions.