The intuitive data interpretation project seeks to transform the binary decision between upfront curation (The ETL approach) and no curation (The NoSQL approach) into a gradient by making ETL processes operate “On-Demand”. An ETL pipeline with On-Demand capabilities can EXPLAIN the effects of deferred processing or curation tasks on query result quality, allowing analysts and domain experts to dynamically balance the need for high-quality information with the costs of obtaining it. Our initial efforts have enabled on-demand source data curation tasks. The next phase of this project will target curation tasks that modify the ETL process itself. As a focusing use-case, we will explore the evolution of processes for extracting structured information from unstructured formats (e.g., log data). As the log format evolves, the extraction process must evolve alongside it. However, requiring the log extraction process to keep pace with the application places undue burden on all participants: the application developers who must communicate changes immediately to ETL designers, the ETL designers who need to react immediately to these changes, and the analysts who are prevented from accessing new data until the ETL designers react. Making such processes on-demand would reduce the burden on ETL designers and application developers, while allowing analysts to continue working even through partial failures of the ETL process.