in the way of my doing business. None of our competitors have
all these onerous requirements.” Rather than simply replying
with a comment about the competition not having a competent
financial crimes officer, we would do well to consider innovative new ways to enhance the customer experience while
meeting regulatory obligations.
Intelligent automation can help reduce the reliance on human
resources. Furthermore, it can help institutions more effectively
mitigate compliance risk, providing for the consistent application
of policies and procedures and a means to organize and accurately
and efficiently review a far greater amount of information.
Understanding Intelligent Automation—
Benefits and Costs
Intelligent automation can be achieved using a number of approaches. These may seem like buzz words now, but they are
rapidly becoming part of the financial crime officer’s vocabulary.
Of course, the old adage “garbage in—garbage out” remains true,
and so does the need to realize that, although there may be a correlation that emerges from intelligent automation, the machine
doesn’t know when a data point shouldn’t be considered.
■ ■ ■ Robotic Process Automation (RPA): RPA can result in a significant reduction of time spent in what are normally highly
repetitive manual tasks. The virtual worker, or “bot,” can complete tasks autonomously within the existing infrastructure by
interfacing with existing internal systems or external sources
to gather and prepare information for human analysis. This
results in a reduction in human activity on low-value tasks,
allowing for humans to focus on higher value, higher-risk tasks.
For example, RPA bots can be used to assist with gathering
information needed for AML alert investigations. The bots can
retrieve customer and counterparty data based upon prescribed
procedures and automatically upload the data into the institution’s case management system. By automating these simple and
repeatable processes, the institution can realize greater efficiency
and streamline their investigative process. RPA bots can also
scan public databases and sources for pending regulations, laws
and rules. Of course, if the RPA bot is not configured correctly,
the results are unreliable and possibly incorrect. That means
there needs to be a validation of the input/output to confirm
accuracy and reliability. It also means the configuration needs
to be regularly visited as data sources will change over time.
■ ■ ■ Machine Learning: Machine learning enables more advanced
aspects of intelligent automation. It represents new capabilities to
automate and augment the analysis of vast amounts of structured
and unstructured data to drive decision making. Models that
gather and consider large volumes of historical data can learn
and begin to predict the likelihood of a particular outcome. In-
stitutions can use Natural Language Processing (NLP) to analyze
unstructured content in documents and interpret the information
to extract meaning and information. This can be done in a frac-
tion of the time it would take a human to process the same data.
And, even if the humans could process this breath of data, lack
of consistency in the outcome is increased and can create risk.
Machine learning can also be especially useful in AML or
sanctions monitoring. Using the historical outcomes of previ-
ous alerts and other data and information available within and
outside an institution (through know your customer (KYC)
data or public source external information), machine
learning models can be trained to identify risks. This
is done by using a pool of existing data as a baseline,
and then learning from feedback provided over
time. Moving forward, this refines the machine’s
understanding of the risks to be identified.
■ ■ ■ Cognitive Automation: Cognitive automation represents the integration of RPA and machine learning
models with advanced reasoning capabilities. This is done by
creating a platform that mimics the patterns of human decision-making while interpreting massive amounts of data—beyond
what is humanly possible. For instance, cognitive automation
can be valuable in uncovering emerging financial crime risks.
Cognitive automation enables an institution to look across the
entire enterprise to identify patterns and risks previously unseen,
while reducing the potential of missing risks inherent in the data.
In addition, cognitive platforms not only retrieve unstructured
investigative content, but, similar to human judgment, they can
also interpret and extract meaning from those sources in order
to help decision alerts with a high degree of confidence. As cognitive platforms acquire new data, they can become capable of
anticipating new problems and modeling possible solutions. This
moves financial crime compliance programs away from typical
rules-based deterministic models into genuine problem solving,
using human-like pattern recognition and hypothesis testing.
Your Journey—Added Value Over Time
Any foray into intelligent automation should not be taken lightly.
Financial crime officers should take into account any lessons
learned from previous “technology waves.” While the benefits of
adopting intelligent automation will outweigh the costs, bankers should be prepared to assess possible pitfalls and develop a
mitigation plan. To make a reasoned decision regarding which
intelligent automation to implement, financial crime stakeholders
need to design an intelligent automation strategy.
First, consider the investment the bank is willing to make, and
the benefits the bank would like to obtain. This process can include
weighing the potential risks with the level of efficiency and agility
the institution seeks. Remember that as an institution moves along
the intelligent automation continuum from RPA to cognitive automation, the rates of return may increase, but so do costs and risks.
For most Institutions, RPA offers the most immediate impact
on efficiency. It is also cheaper and faster to implement. However,
the benefits are limited.
A transparent and easily explainable
intelligent automation framework,
regardless of the level of complexity,
is imperative when planning a strategy to
incorporate within a financial crime