Citi's AI teammate signals a new model for wealth management - Tearsheet
Briefly

Citi's AI teammate signals a new model for wealth management - Tearsheet
Citi launched Citi Sky, an always-on AI-powered member of its Wealth team that engages in real-time conversation and surfaces market, portfolio, and opportunity insights instantly. Clients can talk to it at any time through voice and multimodal interfaces, shifting wealth management from scheduled interactions to continuous availability. The system is positioned with clear guardrails: it does not execute trades, and it interprets, explains, and prepares while human advisors retain final control. The core engineering challenge is controlling non-deterministic generative outputs in regulated settings. The solution relies on strict routing, Citi-specific tool use, tightly bound context, and runtime personalization via controlled data access to preserve relevance, reduce hallucinations, and maintain compliance.
"Citi Sky lets clients talk to it at any time. It delivers real-time market, portfolio, and opportunity insights, shifting wealth management away from scheduled interaction and toward continuous availability. Voice and multimodal interfaces, powered by DeepMind models, push it further into a conversational experience rather than a static dashboard."
"Citi Sky does not execute trades. It interprets, explains, and prepares, while human advisors remain the final point of control. Underneath that product positioning sits a harder engineering problem. For Google DeepMind, the challenge is not building intelligence, but controlling it."
"Generative AI systems are inherently non-deterministic - the same input can produce different outputs, which makes consistency difficult in regulated financial environments. The fix lies in system design, which includes strict routing, Citi-specific tool use, and tightly bound context to ensure the agent operates within defined limits inside Citi's environment."
"Personalization is deliberately kept separate from model reasoning. Instead of being embedded in training, it is applied at runtime through controlled data access, a structure meant to preserve relevance while reducing hallucinations and keeping compliance intact."
Read at Tearsheet
Unable to calculate read time
[
|
]