COSMOS is an AI-powered technical assistant that extracts and assimilates data from heterogeneous sources to accelerate analytics and knowledge discovery. COSMOS can operate over collections of structured (tables), semi-structured (key-value entries), and unstructured data (text, images, and source code) to produce operational data for downstream tasks. To accelerate scientific discovery, COSMOS is currently deployed over xDD (formerly known as GeoDeepDive), one of the world’s largest and fastest-growing collections of scientific publications, housed at the University of Wisconsin–Madison. xDD and COSMOS are backed by high-throughput computing capabilities that provide automation, scalability, and reliability.