Anthropic Executive Urges Banks to Adopt Agentic AI

Banks have largely used generative AI as an assistive tool, with employees entering prompts and models providing responses. Katie McNamara, Anthropic’s head of banking, says the next phase should allow staff to set objectives and permissions, then authorize AI agents to carry out multistep tasks. The shift raises questions about data security, regulatory compliance, accountability and internal governance that will directly affect the pace of adoption.

McNamara told a recent financial summit in New York that AI agents can already operate autonomously for hours at a time. Banks should stop merely “prompting” and start “delegating,” she said. McNamara advocates decentralized AI experimentation across business units under a strict governance framework to manage risk and accelerate transformation. Reports did not disclose the summit’s exact date, the banks in attendance, any investment amounts or a formal implementation timetable.