Arista explains how distributed front-end networks can bring AI inference closer to users, reducing the delays that build up as AI agents make multiple calls to tools and models.