This document presents the specification, architecture, and implemen- tation of a system for identifying and analyzing Politically Exposed Persons (PEPs) built on a Neo4j graph database. The system integrates data on Brazilian politicians, their relatives, business partners, and financial activities, enabling the detection of potential patterns of fraud, corruption, and nepotism. It also incorporates AI components capable of interpreting natural language queries and Cypher queries, and returning contextualized and auditable answers. The primary objective is to provide a robust and transparent tool to support due diligence processes, investigative journalism, and risk analysis in alignment with governance and compliance best practices.