Dun & Bradstreet Restructures Database for AI Agents

Dun & Bradstreet, known for its extensive database of 642 million businesses, has undergone a large-scale restructuring of its infrastructure. Originally designed for analysts who could work with ambiguous data and lengthy queries, the system was ill-equipped for AI agents deployed by clients in credit, procurement, and supply chain management processes, which required a fundamentally different approach.

The issue lay not only in the fragmented architecture but also in the static nature of relationships between companies. For example, traditional systems recorded only current relationships, such as a manager’s affiliation with an organization, but failed to account for dynamic changes—like a top executive moving to another company. For AI agents, such nuances are critical, as they impact real-time risk assessment and decision-making.

Dun & Bradstreet’s solution included consolidating disparate databases in the cloud, creating a unified knowledge graph, and implementing a structured access layer for agents. Special attention was given to an entity verification mechanism: every query passes through a matching system to eliminate identification errors. Additionally, the company developed a "Know Your Agent" concept, similar to KYC, for authenticating machine users and controlling access.

Experts note that Dun & Bradstreet’s experience is relevant for many enterprises. Successful AI agent adoption requires standardized and normalized data, support for dynamic relationships, and built-in mechanisms for entity verification and information provenance tracking. Without these elements, automated systems risk making decisions based on outdated or contradictory data.