Comment by rush86999

Only solution is to train the issue for the next time.

Architecturally focusing on Episodic memory with feedback system.

This training is retrieved next time when something similar happens