Why do we need an AI data scientist? In our combined ~20 years of ML work, we’ve seen the same pattern repeat: in every project, data scientists spend roughly 20% of their time understanding the business context and objectives, 60% manipulating data, and 20% building models. More often than not, that first 20% has by far the greatest impact on the project's success, while the remaining 80% is mostly "undifferentiated heavy lifting": repetitive and formulaic (yet highly technical) work that follows the same patterns across projects. That 80% has two major consequences: Data scientists spend most of their time on repetitive technical work instead of deeply understanding the broader context. The technical know-how prerequisites of that work locks out others (software engineers, analysts, PMs). If we automate the technical but formulaic 80%, we can make data scientists far more productive, as well as open up ML to everyone else. What would you build if technical know-how wasn’t a barrier?