With the introduction of big data analytics and artificial intelligence in customer-facing applications (notable examples are Amazon Alexa, Facebook, Google Search) over the past 15 years, what used to be applications with well-defined behavior has evolved into real-world laboratories at scale. In such an environment, new use cases are discovered after the fact from patterns in user requests, new algorithms and methods are developed on data collected in live customer traffic, and validation and tuning is performed by multi-variate testing against segmented user populations. In this talk, we will review how such an approach relates to a more traditional “function-first” view towards building production-grade software, and will describe some of the key challenges that need to be solved when building such experimentation-first environments. We will conclude by illustrating how we think about incorporating these ideas in our work towards realizing the intelligent hospital at the new SAP Innovation Center in Newport Beach.