This document summarizes the learnings and evolution of EZML, a proposed machine learning tool. It began by targeting data scientists, but interviews revealed that feature extraction was more important than automated model selection. The target then became companies lacking data science capabilities. Further interviews identified the ideal customer as a startup CTO with a recommendation or engagement problem. An MVP was developed with tiered pricing and consulting. Ongoing challenges around data privacy and costs were noted. The document concludes by questioning the business viability and next steps.