The Bank for International Settlements (BIS) recently conducted research into AI agents and their potential for cash management in payment systems. In a report published on November 26, the bank says that Generative AI (GenAI) agents can replicate key carh-management tasks, even without specialized training.
The report comes as banks around the world are spending billions bringing artificial intelligence into their everyday operations. However, amid the AI craze, BIS wanted to see whether GenAI can assist in managing cash and liquidity in real-time gross settlement (RTGS) payment systems.
To determine this, the researchers used prompt-based experiments with ChatGPT’s reasoning model. The goal was to evaluate whether an AI agent can perform high-level intraday liquidity management in a wholesale payment system. The bank stressed that RTGS payment systems are essential for processing large financial transactions between banks in real time, and managing liquidity within them is a delicate balancing act.
The experiment involved simulating payment scenarios with liquidity shocks and competing priorities. The researchers believed this could test the agent’s ability to maintain precautionary liquidity buffers, dynamically prioritize payment under tight constraints, and optimize the trade-offs between liquidity usage and settlement speed.
Research Shows Promising Results For The Future of AI in Banking
They found that the AI agent was able to closely replicate key prudential cash-management practices despite having no domain-specific trading. It also issued calibrated recommendations that were able to preserve liquidity while minimising delays.
Commenting on the findings, BIS said that this suggests that routine cash-management tasks could be automated through the use of general-purpose large language models, and potentially reduce operational costs while improving intraday liquidity efficiency.
“The AI agent was able to maintain precautionary liquidity buffers, prioritise urgent payments, and balance trade-offs between liquidity costs and settlement delays. The agent’s decisions were consistent across various scenarios, demonstrating its ability to adapt to uncertainty and make informed choices,” BIS added.
Researchers are optimistic about the findings, but they also made sure to highlight the need for regulatory safeguards, human oversight, and further research to ensure safe and responsible adoption of AI in financial market infrastructure.
The findings may signal the arrival of even greater AI adoption in modern banking. The bank noted that the research could contribute to the growing discussion on the role of AI in financial systems.
Finally, BIS said: “We conclude with a discussion of the regulatory and policy safeguards that central banks and supervisors may need to consider in an era of AI-driven payment operations.”