Knowledge-Centric Information Systems

This conceptual paper argues that enterprise AI requires a new architectural discipline for organizational knowledge. As knowledge becomes executable infrastructure that systems retrieve, assemble, reason over, and act upon, familiar data-engineering guarantees must be redefined. The paper maps ETL to knowledge ingestion, CDC to knowledge change detection, lineage to provenance, catalogs to knowledge catalogs, materialized views to knowledge views, and medallion architectures to raw-curated-operational knowledge layers. It presents LLM Wiki and the Open Knowledge Format (OKF) as early evidence of this transition, while emphasizing that knowledge architecture is useful only when knowledge becomes an operational asset for humans, agents, workflows, and models.