I meant for a model. Like agents could be many within an organization, instead of patching up each agent with a security solution, wouldn’t it be easier to build one that wraps around the model ?