Does anyone know any prompt engineering to stop the AI from going into platitudes about EVERYTHING???