Many financial institutions globally have started leveraging Robotics Process Automation (RPA) for their Risk and Compliance needs. According to Industry estimates, the overall spend in RPA is going to reach $1bn by 2020 .They also forecast that RPA tools will evolve significantly, and include AI capabilities; the adoption levels for RPA are going to increase to 40% from the current level of 10%. Also it is estimated that approximately 60% of activities that happen in the Risk and Compliance space can be automated through simple automations. Processes that involve greater complexity and require human judgment can also be automated by 15% - 20% through close interactions between BoTs (Intelligent Robots) and human workers.
RPA as a concept and solution is constantly evolving, and many RPA service providers are looking at integrating cognitive capabilities into traditional RPA solutions. Some of the solution providers are expanding their current solutions by including cognitive capabilities such as Machine learning in the RPA tools itself, while some of the vendors are building integration layers which can be used to connect to other digital technologies to enable seamless interactions.
Initially banks were looking at RPA tactically to provide short terms gains in the following areas:
a) Cost efficiencies
b) Automation of simple repetitive tasks
c) Reduction of remediation errors
However, banks are also looking at RPA from a strategic view point to leverage their other digital investments, such as machine learning, NLP (Natural Language Processing) and so on to help them make quicker decisions, get predictive insights, and enable them to spend more time on review.
In terms of cognitive RPA maturity, many financial institutions are still in ideation stages and identifying apt uses cases which can be automated through cognitive RPA.
Apart from traditional RPA benefits such as cost, speed and accuracy, cognitive RPA combines traditional RPA automations with Machine Learning and NLP capabilities and provides great value in analysis and decision making processes. In risk and compliance there are many evident candidates for Cognitive RPA. Some of them currently in focus are:
Customer credit scoring A credit analyst goes through a host of sub-processes before arriving at a customer credit score. In this case, a cognitive RPA solution which combines with traditional RPA and NLP can help quicken the process. For example, while scoring a commercial customer, NLP reads through the customer’s financial information which is supplied by the customer and the public domains for any adverse information. RPA handles the aspects of bringing all this information together and supplies it to the necessary scoring models simultaneously. RPA generates a suggestive decision report which helps the analyst to make quicker judgments in terms of the customer and pricing the loan.
Suspicious transaction resolution Suspicious transactions in banks are generated through the scenarios which are approved by the regulator and defined in AML Transaction Monitoring Engines. According to regulations, if a suspicious transaction is reported it is necessary for banks to analyse them and file a report. The activity is usually handled by the AML investigation team in a bank which spends considerable amount of time in case investigation. The entire process has many manual touch points where traditional RPA, in combination with NLP, can be leveraged to gather preliminary information for the investigator. For example, NLP can screen details on the customer through internal and external sources while RPA can simultaneously gather standard information which typically is gathered by the investigator, such as details of the customer, account, and transaction, and compile a report with NLP inputs for the Investigator.
Data Quality Data quality management has been a traditional pain area for many financial institutions. The data quality resolution process in the risk and compliance process is very time consuming and manual. In this case, a cognitive RPA solution which combines machine learning capabilities and traditional RPA capabilities is being explored by many banks to enable quick remediation of data quality issues. A machine learning training data set can capture all the instances of data quality issues and provide recommended remediations in an automated way to the data quality analyst, and also the training data set can learn from the final decisions taken by the data analyst as well.
Though these are early days for cognitive RPA in risk and compliance, as the RPA solutions mature and integrate more digital capabilities, the adoption in this space is bound to increase as newer risk and compliance areas come up for adoption, which will enable banks to move from rule based automations to intelligent cognitive automations.