Over the past few years, several companies have experimented with AI/ML – deploying a few models in production. However, the challenges that companies face when beginning to use AI are quite different from the obstacles they may encounter when scaling their AI initiatives. In this article, Manasi Vartak (CEO of Verta.ai) describes the processes, people, and tools needed to build and operate hundreds to thousands of models in production. She recommends standardizing processes to develop and operationalize models and adopting tools that support these processes. In addition, she recommends that companies hire data scientists to create models and ML engineers who optimize, package, and implement models in software applications (these personnel can be organized into pods of “Centers of Excellence”). She concludes by suggesting that ML practitioners consider interoperability, user personas, governance, and opportunities for collaboration when evaluating new tools.