Banks could reduce the human effort involved in transaction-monitoring investigations by 30% to 50% by using AI agents for data gathering, analysis and report drafting, while retaining human oversight for higher-risk decisions.
McKinsey says banks could cut the human effort involved in transaction-monitoring investigations by 30% to 50% by using agentic artificial intelligence, a form of software that can perform multistep tasks with limited supervision. The consultancy said the technology can take on data gathering, research, analysis and draft report writing, leaving compliance staff to concentrate on higher-risk cases, judgement calls and final sign-off. In one example, McKinsey said banks using these tools have seen more than a 40% reduction in human processing time, while AI-generated reports were more complete against preset checklists than reports written by analysts. According to McKinsey, the gains come not from replacing investigators but from reorganising the workflow around machines doing the repetitive work first.
The findings add to a broader push by banks to use generative and agentic AI in financial-crime controls, an area where manual review is often slow, expensive and difficult to scale. McKinsey said successful deployments start with a dedicated agentic AI platform and an “AI factory” approach, in which banks standardise how models are built, tested and governed. The firm also said institutions need careful quality controls, traceability and human oversight, because the aim is to augment compliance teams rather than automate away responsibility for suspicious-activity decisions.
McKinsey also warned that traditional “human in the loop” arrangements may not be enough as AI systems are used more widely across financial-crime prevention. Instead, it called for a broader operating model covering use-case design, triage and mitigation, continuous monitoring, governance and the talent and tools needed to supervise the system. The consultancy said leading banks are already devoting substantial resources to quality control and traceability, while also keeping a close eye on costs such as token usage. Even so, it noted that many AI projects remain stuck at the pilot stage because of concerns over security, privacy and compliance, despite strong appetite for further investment.
Disclaimer: This article is intended to inform and educate, not to recommend or endorse any financial product, investment or strategy. Please consider your own financial circumstances and seek professional advice where appropriate before making financial decisions.
