Data interoperability is becoming a practical requirement for companies moving artificial intelligence into production, according to Snowflake Inc. and its Amazon Web Services integration. The goal is to give businesses a governed data layer that supports AI and enterprise workloads without creating more copies of information. Zahir Gadiwan, partner solution engineering leader at Snowflake, said the real challenge in enterprise AI is not finding another model but getting trusted data, business context, governance and scalable infrastructure to work together in one operating model. Gadiwan spoke with theCUBE’s John Furrier during an interview for the AWS Marketplace Series. Traditional data architectures often depend on copying information into separate applications, which becomes expensive and difficult to manage when AI systems need accurate information from several environments. A governed data layer gives services access to information where it already lives while preserving security controls and limiting duplication. Gadiwan explained the architectural shift is from moving data into every engine, app and AI workflow to creating a governed data layer on AWS that multiple services can access in place through open interoperability patterns across storage, catalog, streaming and AI services. The old approach creates latency, cost, data duplication and data trust issues. By separating storage and compute from governance systems, businesses can choose new services without rebuilding their entire data environment.