Automated AML for banks—and why explainability matters

If you want to automate as much of your anti-money laundering (AML) processes as possible to end the painful cycle of hiring and firing, or you’re aware your current AML controls are not dynamic and explainable and it’s finally time to do something about it, this blog post is for you.

It will cover:

  • What automated AML is
  • Two common but problematic ways of automating AML processes
  • One high-quality, sustainable way of automating AML processes
  • The minimum requirements of any automated decision-making platform for AML
  • How to get started

First off, digitally transforming a compliance function can sometimes seem like attempting to service a Boeing 747 engine at 35,000ft.

The reality couldn’t be more different.

New platform technologies, designed for non-technical business users, enable any bank to easily implement transformative change to the automation of their AML processes. And getting set up can be achieved in a matter of weeks.

By the end of this blog post, you will feel more confident that you can recognise the risks and benefits of three different approaches to automated AML, and how to take your first actionable step to automated AML as part of a sustainable digital transformation.

What is automated AML?

Automated AML is the computerisation of processes, decisions and tasks necessary for compliance with AML legislation and regulations across multiple jurisdictions.

Newer, automated AML approaches reduce the cognitive effort required by your people to manage customers and make recurring decisions (such as due diligence assessments and fraud monitoring).

Two common, problematic approaches to automated AML

Here are two approaches to automating AML to avoid.

  1. The legacy, part-automated approach

Banks’ approaches to AML since the post-2008 financial crisis compliance boom have typically involved: 

  • A part-automated process involving multiple pieces of software
  • Use of gated logic steps or linear decision trees in decision making for risk reduction

This approach is not effective because:

  • People require knowledge, training and experience to help them make well-informed risk decisions
  • Risk decisions are made using data that is difficult to interpret, manual to review and unconsolidated 
  • Gated logic/linear decision trees are:
    • Too static to be responsive to human behaviour
    • Too narrow to be able to give an accurate picture of a customer/client and the risk they really present
    • Too difficult to maintain, because regulations get amended over time—one small tweak in regulation can mean hundreds of new branches in a decision tree

  1. Off-the-shelf compliance solutions

Off-the-shelf compliance solutions—a recent emergence in the AML compliance software marketplace—can help speed up specific processes and lift the burden of manual work.

However, these solutions typically mean accepting compromises, including:

  • Having limited rules customisation
  • Having limited integration into the systems (and software licenses) your people already know and use
  • The trade-off of not truly ‘owning’ your compliance function in-house

The high-quality, sustainable way of implementing automated AML

New platform technologies are far better suited to organisations that want their automated AML processes to be sustainable and robust. This is because they enable AML leaders to implement controls that:

  • Scale dynamic, effective decision-making using human-like logic
  • Manage the bulk of recurring decision-making , leaving people to add value where only they truly can (such as challenging know your customer (KYC) checks)
  • Truly own your compliance function in-house

For instance, Rainbird, an intelligent automation platform, helps you meet compliance obligations in a way that:

  • Can easily be adapted to include new requirements
  • Is accurate and stable
  • Scales best practices
  • Aligns with your organisation’s risk parameters
  • Can be scaled across the enterprise for further cost savings

The minimum requirements of automated AML solutions

If you automate your AML processes, you should ensure you can explain the decisions your automated AML is making.

Automating AML means setting up and allowing software to make recurring decisions according to your firm’s risk and compliance policies. Where you are implementing decisions that affect people, it is important to be ready to explain the reasoning underpinning any decision—for instance, if a customer or regulator asks for it.

Leading commentators that work with the FCA: automated systems should have “sufficient interpretability”. And it is stipulated in the ICO’s guidance on GDPR. If a firm uses automated decision making in a way that affects individuals’ circumstances, it should seek to provide those individuals with “meaningful information about the logic involved”. 

How to get started 🚀

You can take your first step towards the sustainable digital transformation of your firm’s compliance by starting small.

For instance, adopting Rainbird’s intelligent automation means integrating one piece of software that provides a fully customisable layer of control to your compliance function. Our team of in-house automation experts can help guide you through your first truly intelligent automation project. 

We recommend starting with a painful problem in your compliance processes—one which, if automated, would immediately relieve tension and free up your team’s workload.  

Rainbird even comes with built-in workflow automation, for easy integration. It is also often integrated with robotic process automation (RPA) providers.

Learn how to automate AML, while remaining compliant with financial and information regulators

Learn how to automate AML, while remaining compliant with financial and information regulators

1 in 2 financial service firms has accelerated machine learning adoption since the COVID-19 pandemic (source: FSTech). But machine learning is not easily explainable.