AI agents have been quietly reshaping the economics of AML operations at banks, FinTechs, payment providers, and other FIs for several years now — and the case for wider adoption has never been stronger.
The bulk of human effort in AML sits within transaction monitoring (TM). Unlike other compliance processes such as customer due diligence, TM demands a higher level of analytical complexity, said Workfusion.
Transactions can carry significant risk when parties, purposes, or links to other transactions are deliberately concealed. One of the most challenging scenarios is layering — where bad actors fragment large financial movements into smaller transactions specifically to evade investigator scrutiny. Complexity multiplies rapidly when multiple parties, cross-border flows, and multiple jurisdictions are involved.
The cost burden of TM is equally significant. Compliance leaders must operate with near-zero tolerance for missed criminal activity, given that regulators can impose highly publicised fines, inflict lasting reputational damage, and introduce costly remediation programmes spanning several years.
WorkFusion’s AI agents are designed to address the back end of the Detection > Alert Generation > Alert Review > Decisioning process. The front end — Detection and Alert Generation — is already well served by existing tooling. The real resource drain lies in the labour-intensive Alert Review and Decisioning stages, and it is here that AI agents step in, enabling compliance teams to operate more efficiently without sacrificing thoroughness.
Ongoing geopolitical tensions and the continuous evolution of international sanctions regimes place further strain on AML and counter-financing of terrorism (CFT) budgets. Identifying which individuals and entities represent sanctions evasion risk is an enormous undertaking. In response, most banks adopt conservative sanctions screening configurations — meaning almost every transaction triggers an alert requiring manual review and decisioning. As with TM, false-positive rates of around 95% are commonplace.
AI agents offer a compelling solution here too. By taking on the burden of routine alert triage, they free compliance staff to concentrate on genuinely high-risk cases. The downstream benefits extend beyond cost savings: FIs can eliminate backlogs, accelerate payment processing, improve customer onboarding experiences, and ultimately drive broader business growth.
In the area of know your customer (KYC) operations, AI agents bring together a range of AI capabilities — including machine learning (ML), natural language processing (NLP), optical character recognition (OCR), and generative AI (GenAI) — to streamline adverse media reviews.
KYC work encompasses suspicious activity and sanctions detection, investigation, and reporting. It involves five core activities: gathering customer records and third-party data; organising that information for operational teams; assessing relationships and patterns; reasoning through those findings; and making a judgement on whether a matter warrants escalation or can be closed. AI agents perform all five steps by accessing data from standard adverse media screening tools, evaluating relevance, demographic information, material significance, and other attributes. Crucially, agents built with deep knowledge of financial crime can conduct a comprehensive investigation using as little as a name. The results are then surfaced to a human team member for final review.
WorkFusion’s AI agent, named Evan, also produces a detailed, written, and auditable justification of each risk evaluation — a critical feature for regulatory accountability.
A typical adverse media search returns five or more results, with each one taking a human analyst between 10 and 20 minutes to read and assess. Evan processes the same results in just two to three minutes, reducing review times by 80–90%. Through human-in-the-loop technology, the riskiest cases are automatically escalated to team members, ensuring that skilled investigators are deployed where they add the most value rather than being tied up in low-value triage work.
Whether it is TM alert reviews, sanctions screening, or adverse media analysis, AI agents are becoming an indispensable colleague for AML analysts and investigators — absorbing the tedious, high-volume workload that inflates budgets, and freeing human expertise for the complex, high-stakes investigations that genuinely demand it.
Copyright © 2026 RegTech Analyst
Copyright © 2018 RegTech Analyst
