Your data is never pooled into a shared model and never used to improve anyone else's agents.
What we do is post-train a model on your team's decisions for your team alone. It lives in your environment, it learns only from you, and it works only for you.
The model your team's decisions built, and everything it learned, are yours to keep.