Phoenix Group
AI Agent Engineering

// OBJECTIVE

Phoenix Group needed to improve the speed, quality, and consistency of group risk documentation across multiple business units — a process that was manual, time-intensive, and vulnerable to inconsistency.

// THE BREAKTHROUGH

We ingested a large volume of background information about Phoenix Group and built a series of agents that made sense of it — constructing a structured knowledge base that gave every downstream agent deep company context. On top of this foundation, we built a suite of agents that work across the full risk documentation lifecycle.

// APPROACH

The knowledge agents ingested company documents, policies, and historical risk records to build a structured understanding of Phoenix Group that the working agents could draw on. The working agents then read meeting transcripts and construct risk definitions by understanding risk discussions and augmenting them with this broader company context. They complete risk control definitions, audit documentation for quality, consistency, adherence to audit language, and specificity. Across the group, agents check how risks are implemented in different companies, identify inconsistencies, and flag gaps.

// OUTCOME

Risk documentation that previously took weeks of manual effort is now produced faster and to a higher, more consistent standard. The agents serve as a quality layer across the entire risk function — catching inconsistencies that human review alone would miss.