Question Can you train agents in a build?

A friend and I were discussing ML Agents and one of the things that came up was whether or not ai made with ml agents could adapt to someone's playstyle. I'm unsure if ML agents can be used in a build, but tbh only because I haven't seen it done yet. Is this something that's possible? How does that work? How do you handle the parts of the process that involve python?

 

https://github.com/Unity-Technologi..._docs/docs/Learning-Environment-Executable.md