No I'm saying this entire problem approach, including memory system of agents, is incorrect. Like the author argues. With the same solution of moving the ground truths outside of the model/context.