Gartner research shows only 53% of projects make it from artificial intelligence (AI) prototypes to production. There are two reasons for that: First, in the midst of frequently overhyped expectations, a clear path toward the real value for the organization is often not defined for the initial project. The second reason, which is even more important and often ignored: The technical gap between a shiny prototype and putting the results of that prototype into production is big. Bridging that gap between the creation of a combination of data wrangling and model optimization through to deploying that process often requires a complex, sometimes even manual step. Worse, the technologies used are seldom aligned well.

Systems that automate the data science cycle have been gaining a lot of attention recently. Similar to smart home assistant systems however, automating data science for business users only works for well-defined tasks. We do not expect home assistants to have truly deep conversations about changing topics. In fact, the most successful systems restrict the types of possible interactions heavily and cannot deal with vaguely defined topics. Real data science problems are similarly vaguely defined: only an interactive exchange between the business analysts and the data analysts can guide the analysis in a new, useful direction, potentially sparking interesting new insights and further sharpening the analysis. Therefore, as soon as we leave the realm of completely automated data science sandboxes, the challenge lies in allowing data scientists to build interactive systems, interactively assisting the business analyst in her quest to find new insights in data and predict future outcomes.