I guess you have to know which experts to keep “hot” in ram, which you can’t know beforehand, so there wouldn’t be much gain.
Maybe you could even do something like profiling and remember over multiple sessions (per project/workspace) what the most used agents are and load those up before hand.