Financial institutions are increasingly using automation to manage growing AML workloads, but RegTech provider ZIGRAM argues that more automation does not necessarily mean better compliance.
The firm says the focus should instead be on deciding which activities technology can handle efficiently and where investigators need to remain responsible for interpreting risk. The objective is to remove repetitive work so compliance professionals have more time for complex investigations and decisions.
AML analysts can spend significant time on tasks that sit around an investigation rather than on the investigation itself. Gathering transaction histories, reviewing previous alerts, switching between screening and monitoring systems and reconstructing customer context can all add to the workload.
As transaction volumes increase, these processes can contribute to alert backlogs and false-positive fatigue. ZIGRAM argues that this can leave experienced investigators spending too much time on administrative work instead of assessing potentially suspicious activity.
A useful starting point for determining where automation belongs is the combination of high volume, high repeatability and low ambiguity. Activities that follow the same process repeatedly are generally stronger candidates for automation than decisions requiring significant interpretation.
Data retrieval, alert enrichment, screening comparisons, case creation, workflow routing, reminders, approvals and audit trails can therefore be automated without necessarily requiring sophisticated AI. Rule-based workflows can often handle these tasks while providing investigators with a more consistent starting point.
Screening demonstrates where automation has a clear role but also has limitations. Technology can compare large numbers of records against sanctions, PEP and watchlist databases, account for variations in names and identify potential matches.
A potential match does not, however, establish that a customer is the person or entity identified on a watchlist. Incomplete identifiers, transliteration differences and common names can all require additional investigation.
Transaction monitoring presents a similar challenge. Automated systems can process large volumes of transactions and identify unusual patterns across values, frequencies, geographies, counterparties and customer behaviour.
The Basel Committee has indicated that effective monitoring at most banks, particularly internationally active institutions, is likely to require automation. However, unusual activity is not necessarily suspicious activity.
A rise in cross-border payments, for example, could represent increased risk or reflect a legitimate expansion into a new market. Automation can identify the change, but an investigator needs to establish why it occurred and whether the explanation is credible.
ZIGRAM places several activities between full automation and manual investigation. Risk scoring, alert prioritisation and network analysis can all benefit from technology while retaining human oversight.
Risk scores can combine customer, transaction and behavioural information to highlight potentially higher-risk cases. For investigators, however, the score needs to be explainable. They should be able to understand what caused it to change, which factors contributed to the result and what action should follow.
Network analysis can similarly identify relationships between customers, accounts, counterparties and beneficial owners at a scale that would be difficult to replicate manually. Investigators still need to determine whether those relationships represent legitimate activity or warrant further scrutiny.
The need for human accountability increases when AML decisions become more complex or consequential. Enhanced due diligence, material escalations and SAR or STR decisions can involve questions around source of funds, source of wealth, ownership structures and broader customer context.
Technology can gather evidence, organise information and support these processes, but the final reasoning needs to be defensible. As ZIGRAM puts it, “The model gave it a high score” is not enough to explain a compliance decision.
The firm also identifies risks associated with poorly designed automation. Weak data can be processed faster without being corrected, while excessive reliance on system recommendations can create automation bias among investigators.
Explainability is another consideration. FATF has highlighted challenges around the interpretability and governance of new technologies used in AML and countering the financing of terrorism.
There is also a risk that institutions automate individual processes without connecting them. Screening, risk scoring, transaction monitoring and case management can each become automated while investigators continue to manually transfer information between separate systems.
ZIGRAM argues that the next stage of AML technology should instead focus on connected decision-making. Customer risk assessments can inform monitoring scenarios, screening results can update customer profiles and investigation outcomes can feed into future risk assessments.
The company’s Complete AML System follows this approach by connecting customer risk rating, watchlist screening, transaction monitoring and case management within a broader AML workflow.
For compliance teams, the distinction is between automating individual tasks and creating a more connected operating model. Technology can handle predictable activities, support analytical work and provide investigators with more relevant information, while humans retain responsibility for decisions that depend on context and judgement.
The result is an AML model focused not on automating the greatest possible number of processes, but on using technology where it can improve efficiency without removing the oversight required for complex financial crime decisions.
Copyright © 2026 RegTech Analyst
Copyright © 2026 RegTech Analyst
