Visual analysis helps people see and understand data. Effective visualizations depend on the task at hand and need to be simple, yet meaningful. While data-driven inquiry has become the norm of business practices and decision making, there is a huge untapped market of “data enthusiasts” who aren’t database or computer experts; yet, they are great analytical thinkers and need tools to support their questions. There have been recent advances looking at how AI technologies can assist the analytical workflow ranging from smarter data transformations, automatic visual encodings, to supporting analytical conversation using natural language. Machine learning approaches have shown to be promising for approximating the cues for continuous learning in these systems. With a better understanding as to how users explore data in their flow of analysis, a natural question is - can user behavior be applied as a set of engineering requirements to developing smarter tools? That is, can people doing analysis be supported or even replaced by more intelligent tools? In this talk, I will explore this question.