Kelsey talks with CIT President and CEO Kyle Etter about recent AI releases like GPT 5.4’s native computer control, Gemini 3.1 updates, and the shift from chatbots to agentic AI that can navigate business systems and complete real work. They discuss why IT infrastructure must now include AI governance, identity, logging, permissions, and emerging liability and insurance considerations for “digital teammates,” noting early examples like insured voice agents. Kyle advises deploying narrow, task-specific agents with strong guardrails, human-in-the-loop checks, and layered QA/security agents to reduce risk. They also cover preventing shadow AI by providing authorized business licenses, training staff to manage agents, and why ROI is increasingly driven by integrating agents into existing software and data workflows, including analysis and data cleanu ...