AI Agents Move From Experimentation to Real-World Use in Financial Services

Financial-services companies have traditionally used AI mainly to compile and analyze data, while employees still gather information across systems, review documents and navigate multiple approval layers. Snowflake, KPMG and The Economist say AI agents can connect data and workflows in governed environments. Initial deployments are targeting repetitive tasks such as KYC, audits, fraud detection and mortgage reviews, signaling that AI is moving from offering recommendations to taking action.

Snowflake introduced its Ecosystem Agent Framework on April 22, 2026, and PYMNTS summarized related developments on April 27. A KPMG survey found that deployment among large institutions rose from 11% in the first quarter of 2025 to 26% in the fourth quarter. Oracle simulated a mortgage workflow spanning 45 procedures and estimated that approval times could fall from 48 days to 38 days, while completed application volumes could rise 13%. However, none of the parties has disclosed actual investment amounts.