Tamr is a new approach to data integration that uses supervised machine learning to make it faster and easier to unify large numbers of heterogeneous sources. Unlike traditional rules-centric approaches common in ETL and MDM products, Tamr enables data engineers to leverage machine learning (enriched by the expertise of humans who best understand the data and supplemented by rules where appropriate) to rapidly construct a probabilistic data unification model that combines, masters, and classifies data from myriad siloed sources. The result is a new class of data assets that deliver a unified view of key entities (e.g. customers, products, suppliers, parts, etc.) and power breakthrough analytic insights and operational improvements.