I will study it.
Although I already have the behavior tree of my AI done.
I spent a lot of time doing it.
I will use it if I do another type of AI (just to learn something new and compare between both methods).
What do you think? Which of the two is more efficient?
You already know that I am in the process of optimization (for two months now).
Well, I’ve gotten around 80FPS on average in the editor (Min 60 - Max 100).
My goal is to reach 120FPS in the editor… but it’s complicated
Hope the op doesn’t mind the slight derail. They can consider their options in regards to future optimizations.
Thank you 
@br_chickn Whenever you want to optimize you can ask me whenever you want… just send me a message
Same topic, different scenario.
All my bots are placed in the level. They are not spawned from a spawner. The soft reference array that loads the tasks is exposed. This is only for movement task like search, flank and such. These task will be loaded through the construction script. When the exposed array is altered, all the tasks are destroyed then loaded back in. Will this cause problems as the load/unload is not happening at runtime?
I guess I could compare the old task array to the new array and change the differences
In our private messages i explained you how to make “a central AI manager” as @3dRaven says (I hope you will read it)
if you get a invalid reference in the constructor script you will never can open de blueprint again (it is just a warning). That was my first disaster with blueprints. And that was the day i decided do it all in C++. That broke my all my project.
I thought you were using ActorComponents
What kind of components are you using?
