Compliance screening increasingly depends on information that is distributed across public institutional websites, documents, and other external sources. The practical challenge is not only discovering this information, but acquiring it reliably, preserving its provenance, transforming it into a consistent representation, and maintaining an auditable path from a normalized record back to the source from which it was obtained. This paper presents the design and current implementation of the Compliance Intelligence Platform (CIP), a Python-based modular ingestion system intended to support compliance-oriented analysis of politically exposed persons and related institutional relationships. The platform separates source collection, raw-document preservation, parsing, normalization, entity modeling, relationship resolution, persistence, and audit tracking. Its data model represents persons, positions, institutions, source documents, and person-position relationships, while the ingestion pipeline records source runs independently from the main data transaction. The initial source-specific investigation focuses on publicly available National Assembly information from the Parliament of Kenya and uses browser rendering and immutable HTML snapshots to address dynamic web presentation and reproducibility concerns. The current implementation establishes the core architecture and test infrastructure but remains an active development project: the National Assembly adapter currently fails to discover pagination reliably in the tested live-rendered response, and several downstream extraction and production integration stages remain incomplete. The paper therefore reports an engineering design and implementation study rather than claiming a production-ready compliance screening product. The work demonstrates how provenance, deterministic source snapshots, separation of concerns, transaction-aware persistence, and explicit validation can form a maintainable foundation for a multi-source compliance intelligence system.