Aghi Marietti tells Sophia Dew that AI agents create new infrastructure security concerns. He describes gatewaysA large language model gateway is an intermediary that standardizes, routes and monitors application calls to one or more language-model providers. as checkpoints for observing, throttling or blocking requests, and distinguishes model token traffic from agents executing API calls.
Marietti describes enterprise adoption of multiple models and says token use is rising in the traffic his company can observe. His figures and forecasts describe that subset, not the entire market.
They discuss the difficulty of connecting token spend to useful outcomesCost per completed task measures the total AI, tool, infrastructure, retry, and repair expense for each verified useful outcome., as well as the role of budgets, agent identityAn AI agent identity is a persistent, verifiable identifier used to attribute messages, actions, permissions, and history to an agent. and registries in managing enterprise AI workflows.
Watch on YouTube
