Business applications are usually data-intensive. The process of designing and implementing such applications benefits from working with realistic data to sharpen requirements and discover pitfalls. However, such data is usually quite extensive and the feedback cycles during programming and design activities can become long and distracting. As a result, programmers might prefer abstract thinking and mental simulations over working with concrete, realistic data. We propose a new approach supporting live programming, with immediate feedback and explorable runtime data, for the domain of data-intensive business applications on top of relational databases. With the integration of streamed access to sampled data, we can employ productive traits of a live programming environment such as Squeak/Smalltalk, which is not optimized for the processing of huge amounts of data and is hence not well-suited for such tasks. We describe two representative scenarios and also discuss limitations by putting our approach in relation to the current development of business applications.