Machine Learning Patterns - The First Touch or Four Piece?

A lot of data, machine learning engineers and data scientists haven't come into the industry through the software engineering route. So we don't know about a lot of stuff that really exists. If you go into software engineering heavily enough and I'm learning this stuff, there's a load of tools for this already. For me, it's like saying the same thing over and over again but what's the smallest thing you can deliver to increment what they're doing? Where do you want to get this all line up with a data strategy and it should all end up with text Strategy which is loads of chat,. But what's the small step that gets us just on the next rung in the ladder