Data mapping is a crucial step during application migration and for application integration. Data (model) mapping comprises “schema matching” to identify semantically equivalent fields between two schema, and “transformation logic generation” to write rules for converting data from one schema to the other. In industry practice today, data mapping is largely manual in nature, done by domain experts. We present a data mapping assistant powered by Large Language Models (LLMs), providing disruptive precision improvement over SOTA methods and multiple automation workflows that let users provide different available input triggers (context) for inferring the mappings. We illustrate the contribution using various representative industrial datasets.