Manual processes are prone to human error. This can result in long, drawn-out processes which can be extremely costly to an organisation. AI could hold the key to automating these processes though.
Harinder Singh Sudan is senior vice president, financial intelligence unit at BlackSwan Technologies. BlackSwan Technologies is reinventing enterprise software through agile intelligence for the enterprise – a fusion of data, AI and cloud technologies.
With close to 20 years of industry experience in banking and financial services, Singh Sudan leads BlackSwan’s FIU practice globally.
Singh Sudan has worked with a number of tier one banks and consultancies. He has also managed service providers in the UK, Europe and Middle East. He has deep expertise in financial crime compliance; covering operations, technology architecture, program management and delivery of global remediation programs.
Speaking to The Fintech Times, Singh Sudan explains how AI can tackle challenges that have previously challenged compliance due diligence:
For a number of years, firms have sought to modernise their financial crime compliance systems. One of the key aims is to reduce the amount of manual work carried out by analysts for KYC, Screening and Transaction Monitoring. Ultimately, it aims to do a more effective job of complying with regulations and avoiding crippling regulatory fines.
Subsequently, firms would be able to increase the efficiency of processes, reduce case processing time and reduce costs. And yet, organisations are still facing a number of challenges in achieving this aim.
Ineffective manual searches
For instance, KYC analysts are tasked with reviewing documentation for new customer accounts, evaluating high-risk accounts and gathering due diligence information. The open-ended searches are where they look for external information on an entity. They collect everything they can find from a large number of predefined sources.
These manual searches are inefficient. Not only are they time-consuming but they make it difficult for analysts to determine which information is of the highest quality and therefore understand how to prioritise it.
The poor data quality leads to a high number of false positives. This makes it challenging to create a reliable and comprehensive customer profile. And this problem is compounded as the data search can’t be audited or reproduced. Therefore, if there is a change to a client’s profile which could be significant, firms would most likely miss it for a long period of time. This increases their risk exposure and therefore their likeliness of falling foul of the regulations and paying a fine.
Screening woes
The next step of the financial crime compliance process is screening. Each stakeholder – whether it be the CEO or a shareholder – that had been identified during the open-ended searches during KYC, has to be screened against multiple sanctions lists, adverse media files and illicit activities.
The remaining challenge for many firms here is that they are still reliant on analysts screening one entity after another against specific criteria, which can be painstakingly slow. Particularly as firms often want to expand the screening to clients, business partners and internal staff.
Banks are also tasked with analysing transactions generated. They must ensure that anything going through the organisation’s system is clean. But with money laundering booming, firms are being exposed to more fines, and this is because their existing set-up is made up of a battalion of analysts in the back office sifting through each and every transaction.
With the increasing use of real-time payment systems, this becomes even more of an issue. When a new transaction is made, banks have to identify if the sender or receiver is on a watchlist. If the country or countries in which the transaction is made are high-risk, and if the amount sent meets the approved threshold. This manual approach is tedious and prone to errors.
A proportion of analysts’ time is spent on mundane tasks, one of which is maintaining the workflow system in place. Most organisations cross tasks off a to-do-list manually in a spreadsheet, so if every stage of the workflow has five stages, each stage consists of a to-do-list. This reduces the visibility of work in progress and means there is no way of initiating feedback.
Automating the process
There are certain processes which firms should focus on automating in order to get the best results. This not only saves time but increases the reliability and accuracy of the process. It enables the analyst to perform the job they were hired to do – which is to analyse the data that’s in front of them.
For KYC, automation begins at the questionnaire stage; rather than going through 100 questions one by one. These can be automatically filled using the searches about an entity. The searches themselves can be automated, so that the KYC analyst does not have to perform each one separately.
Profile building
Crucially, a proactive profile builder can feed the results of these searches into a customer profile. The search process itself relies on premium sources to build the organisational structure of the target. Therefore, the organisation can identify the target they’re investigating. But also who owns it, which is usually a very time-consuming process. However, the organisation can reproduce the same technique multiple times to eventually find out who the Ultimate Beneficial Owner (UBO) is.
As KYC is only carried out once or over regular intervals, there is a chance that organisations can miss critical updates to a profile. A solution that automatically monitors and updates a profile when there are any changes to a client is key. The automation means that even if the firm has 20,000 clients or a million clients, they can ensure all of the profiles are up-to-date without the need of manual intervention.
Firms often struggle to address false positives when it comes to KYC. However, sophisticated algorithms can lower the number of false positives. Fuzzy logic, for example, allows users to set a confidence level of what they want to find. This means users can set parameters that say ‘I want to find exactly x’ or that confirm they are happy with five per cent of false positives. In some cases, if it’s likely to be a popular name in the search, they would be happy with a higher proportion of false positives.
Replacing old with the new
By replicating what the algorithm does and by changing its parameters, firms can adapt to a new risk appetite. This isn’t something that can be carried out with an army of analysts. This is because each analyst has a different competence and all analysts have a fatigue threshold. It’s difficult for a firm to assess the ability of an analyst at the time of determining a true or false positive.
By carrying this out on a very deterministic basis, compliance teams can have the confidence of the model’s accuracy at all times. And crucially, the process can be reproduced.
Integration between the different financial crime compliance modules is key to streamlining the process further. The list of names of all the stakeholders of an organisation that have been identified during the KYC process can then be automatically screened against all watchlists and other criteria.
This replaces the existing method for many firms of screening one entity at a time against each list. When a hit is generated during screening, the system orchestrates the way you deal with each one until this has been completed, then the case can move to the next task in the queue.
Standardising a fragmented process
This fully digital workflow replaces spreadsheets and different analysts carrying out tasks in their own distinct ways. The benefit is that this brings an element of standardisation to the processes within the company. In cases of financial crime, this is paramount as these are real investigations on people behind a company. It needs to be auditable, reproducible and documented.
This is also true of the Transaction Monitoring process – by replacing the existing manual methods with a sophisticated, real-time transaction monitoring engine, organisations can automatically distinguish if a transaction’s sender or receiver appears on any lists, is sending the money from a high-risk country, is sending an amount of money according to the approved threshold, and numerous other factors that would take analysts or less sophisticated systems far longer.
Regulators look positively towards modern tech
A solution which has automated search functionalities, profile building and transaction analysis will help organisations to keep on top of financial crime by reducing the reliance on manual methods. Meanwhile, a streamlined digital end-to-end process for the client is more cost-efficient. It also enables analysts to be more proactive in how they investigate hits.
An automated, AI-enabled application would signal to the regulator that the organisation is using modern technologies. It would also show they are committed to be part of the fight against financial crime.
