Home / Technology / D&B Rebuilds Data Graph for AI Agents
D&B Rebuilds Data Graph for AI Agents
22 May
Summary
- Dun & Bradstreet rebuilt its Commercial Graph to serve AI agents.
- The original graph architecture was unsuitable for machine queries.
- New systems address agent identity and data verification challenges.
Dun & Bradstreet has completed a major reconstruction of its Commercial Graph, a vast commercial database with over 180 years of history. This overhaul was driven by the emergence of AI agents as a new consumer category, requiring a data architecture fundamentally different from that serving human analysts. The previous system, a collection of fragmented databases, struggled with the scale and dynamic relationship tracking needed by machines.
The company consolidated its data onto cloud infrastructure and developed a unified knowledge graph. This new structure normalizes records across markets while adhering to regional compliance. To facilitate AI interaction, D&B created a structured access layer with tools to package data contextually and ensure accurate entity resolution.
Addressing AI agent identity, D&B established a "Know Your Agent" model, akin to "Know Your Customer," requiring IP address mapping and unique access keys for authentication. Furthermore, a verification agent was developed to maintain entity consistency across multi-agent workflows, ensuring that chained agents remain focused on the same business entity throughout their processes.
