Banks Have Shifted to Autonomous AI Agent Workflows
Financial institutions are moving beyond copilots, tasking autonomous agents with complex payment and trade processes.
Updated on Sept. 29, 2026 in Financial Services
Live Poll
Do you believe AI should fully automate workplace tasks rather than just assisting human judgment?
Banking institutions have transitioned from using basic AI copilots to autonomous agentic workflows to handle labor-intensive financial tasks. These systems, designed to improve operational efficiency and reduce manual intervention, are already being deployed for tasks like payment repair and trade processing.
Why it matters
Financial institutions are adopting these agents to eliminate the overhead of manual data intervention and management infrastructure. This shift aims to automate complex assessments and free up human staff to focus on higher-value financial activities.
BNP Paribas trade processing agents reached 80-85% efficiency after training compared to previous manual benchmarks. Google manages $170 billion in total assets—including a $100 billion portfolio—across 60 countries and 100,000 daily transactions using automated treasury systems.
The players
BNY
A global financial services firm providing investment services and management, currently testing digital employee agents.
BNP Paribas
A major international banking group that has integrated AI agents into its trade processing infrastructure.
A global technology conglomerate that functions as a sophisticated treasury operator managing $170 billion in assets.
Deutsche Bank
A global investment bank currently exploring the deployment of multi-purpose agents for complex financial assessments.
The details
Banks are integrating guardrails into deterministic AI models to ensure verification and oversight remains in the loop for sensitive transactions. Google’s treasury system exemplifies this architecture, utilizing separate forecasting and evaluation agents linked directly to bank connectivity. These agents stage and execute trades automatically, a departure from traditional manual execution models.
Timeline
September 2026: AI developments were presented at the Sibos 2026 conference.
Market Landscape
The push toward autonomous agentic workflows marks a significant evolution beyond the first generation of AI copilots showcased at the Sibos 2026 conference. This shift follows a broader industry trend of replacing manual, human-intensive payment repair with deterministic AI systems.
Operators should evaluate their own manual financial workflows for potential conversion to autonomous agent systems to improve margin. Compliance and legal counsel must be consulted to ensure human oversight guardrails are sufficient for automated high-value transaction execution.
The takeaway
The move from copilots to agents highlights a critical transition from simple assistant tools to automated operational workers. Monitor the efficiency gains reported by major banks like BNP Paribas to determine when your own infrastructure reach warrants similar AI implementation.
Further reading
For more information on current industry trends, see the Financial Services section.
Source note: This article includes information reported by Finextra Research.
Live Poll
Do you believe AI should fully automate workplace tasks rather than just assisting human judgment?
