Banco Popular Dominicano: AURA agents for operational risk

Operational risk was reviewed periodically and by sampling. Teams analyzed documents and validated controls by hand in cycles that could take months, and Microsoft's story says this covered only about 40% of the bank's risk universe.¹

AURA is a multi-agent system built on Microsoft Copilot Studio and Microsoft Power Platform. It combines document agents, risk analysis agents and conversational agents, with supervisory agents that oversee the rest. Business experts can create and adjust agents without waiting for technical teams. Power Automate starts analyses when risk matrices, documents or controls change, Power BI tracks system stability and the health of the risk pipeline, staff work with AURA through Microsoft Teams, and it connects to Excel, SQL Server and SharePoint.

AURA proposes and analysts decide. Access control, separate roles and logging of every process and result let the bank audit each step. Local press reports that the bank's own risk management teams designed, tuned and scaled the system. Microsoft describes it as an established operating model rather than a pilot, still being extended to more areas, users and processes. In June 2026 AURA won three Latam Digital 2026 awards, including first place for integral digital strategy.¹²

In our view, the notable part is who built it: a regulated control function, not the IT department, assembled a supervised multi-agent system on low-code tools. Other banks should copy its controls first, with analysts making every decision, separate roles and full logging, before they aim for its coverage figures. The numbers come from Microsoft and the bank, carry no named source or method, and the accuracy measure is not defined, so they show direction rather than a benchmark. A bank with a small, stable set of controls may get most of the benefit from Power Automate flows and Power BI reporting without agents.