Citigroup will test autonomous artificial intelligence capabilities that can complete multiple tasks through a single user prompt across various company systems.
The Wall Street Journal reports that the banking giant is implementing new “agentic” AI functionality within its proprietary platform, marking a significant step toward technology that could autonomously handle complex assignments similar to those of human workers.
The pilot programme spans 5,000 users over four to six weeks, testing capabilities within Citi Stylus Workspaces. The system enables staff to direct AI tools to research specific clients, build comprehensive profiles using both public and internal data sources, and translate results into foreign languages through single instructions.
Chief Technology Officer David Griffiths explained that whilst similar functions existed previously, they required separate human direction at each process stage. The enhanced platform utilises multiple AI models, including Google’s Gemini and Anthropic’s Claude.
“A couple of years ago, you could do agentic things with the early versions of the models that were available then. But they weren’t always very reliable. They weren’t always very good at invoking tools. But they are now,” he said.
The technology addresses thousands of similar workflow challenges throughout the organisation, according to Griffiths, who will monitor usage patterns, impact measurements and cost-value ratios during the trial period.
Cost management remains crucial as agent operations can become expensive rapidly, particularly for complex, extended tasks. The company has implemented hard cost limits within systems, though declining model usage costs complicate accurate return-on-investment calculations.
Current tasks are completed within minutes, but longer-duration operations could alter workforce considerations significantly.
“Does it mean that we need less people? I don’t know,” said Griffiths. “It certainly means that we would get a lot more done. And we’ll see how the workforce evolves with that massive boost of capacity that we’re getting here.”
The initiative reflects broader corporate investment in AI agents capable of handling everything from grocery shopping to application coding, with companies exploring how multiple agents can communicate safely with each other and internal systems.
