As a researcher in program verification and static analysis, I was an admirer of the elegant theoretical work on Abstract Interpretation. Then, in 2013 I moved to industry and the shock of attempting to deploy program analysis to people made me appreciate the subject much, much more. Going back and forth between the theory and the needs of people who use the tools made me revisit my own assumptions, leading to surprising (for me) places. In the talk I’d like to tell you you what I learned about AI and how it made a difference, as well as what I wish I knew and why there is much to do.