AI agents are becoming an increasingly important tool in financial crime compliance, helping banks and other financial institutions automate complex tasks and reduce risk across AML programmes.
While automation in compliance is not new, the technology underpinning these agents has significantly advanced, and many of the world’s largest banks have relied on digital workers in their AML operations for several years, claims Workfusion.
However, as interest in artificial intelligence grows, so has confusion about what AI agents actually are, prompting FinTech vendors to apply the term so liberally that it has blurred its meaning across the market.
Many compliance and technology professionals have found themselves unsure whether AI agents are simply rebranded bots, a new way to describe generative AI, or an extension of large language model capabilities. To help clarify the technology and its role in modern compliance functions, WorkFusion’s chief technology officer Peter Cousins, VP of financial crime compliance David Caruso, and VP of AI solutions Kyle Hoback recently discussed how AI agents differ from other tools and why they are essential in the evolving fight against financial crime.
Peter explained that bots and AI agents are often mistakenly grouped together, even though they operate very differently. Bots are useful for highly structured and repeatable tasks, but they lack the flexibility and contextual reasoning needed in FCC workloads. When asked about this distinction, Peter said, “We usually bristle a little bit when people refer to our AI Agents as bots, because they are so much more than that.” He noted that bots are fast and reliable for locked-down tasks, whereas AI agents mimic human judgement and handle more fluid work patterns.
He also warned against assuming AI agents are interchangeable with large language models or generative AI. While some agents may incorporate these AI elements, their role in compliance requires a greater focus on explainability and controlled outputs. In regulated environments, reliability and consistency are essential, and decision-making must be transparent.
David highlighted how crucial this level of explainability has become as regulators increasingly expect banks to deploy AI-driven technology. WorkFusion AI agents are already in use at 10 of the top 20 banks worldwide, and he said, “You have to provide all of the documentation that explains how these agents work.” He emphasised that without oversight and clarity, regulators will not allow AI agents to operate in critical compliance processes.
AI agents, despite their name, do not rely on AI for every step in an AML process. Peter noted that many stages of FCC workflows benefit from intelligent automation without requiring advanced AI techniques, although tools such as machine learning and optical character recognition are commonly used. Over time, institutions can expand their use of AI agents into more complex areas, such as adjudicating alerts and performing enhanced due diligence. The recommended approach is to start small, measure improvements, and scale as confidence grows.
As regulators and financial institutions continue to adapt to more sophisticated criminal threats, AI agents are rapidly becoming a foundational capability in FCC operations. The question from supervisors is no longer whether firms should use AI, but as Peter put it, whether they are “sure you don’t want to use AI in such an important process?”
Copyright © 2025 RegTech Analyst
Copyright © 2018 RegTech Analyst
