The secondary use of clinical data for biomedical research is a cornerstone of modern medicine, yet its potential is severely constrained by the persistent challenge of data heterogeneity. While the Observational Medical Outcomes Partnership Common Data Model (OMOP-CDM) provides a robust standard for data harmonization, the process of transforming source data into this model remains a significant bottleneck, characterized by manual, resource-intensive, and brittle Extract-Transform-Load (ETL) pipelines. This paper proposes a novel agentic architecture that reframes data standardization from an imperative scripting task to a declarative, AI-augmented knowledge-generation process. Our framework decomposes the complex mapping workflow into a series of discrete tasks executed by specialized, autonomous software agents. This modular architecture orchestrates a powerful synergy between declarative data modeling with the Linked Data Modeling Language (LinkML), reliable Large Language Model (LLM) interfacing via the Boundary AI Markup Language (BAML), and the cognitive capabilities of LLMs for model alignment. The system's primary output is not merely transformed data, but a human-readable and machine-executable mapping specification using linkml-map. This approach establishes a new paradigm for creating scalable, reproducible, and transparent data standardization pipelines, positioning the data engineer as an expert validator of AI-generated knowledge rather than a manual coder of transformation logic.