Applying Artificial Intelligence in Wealth Management: Compelling Use Cases Across the Client Life Cycle. This is part of a series of articles drawn from a major report of this title that is produced by this news service.

Wendy Spires, Head of Research at WealthBriefing, discusses how the “digital detective work” AI can carry out will make client due diligence a very much more efficient – and far less risky – affair for wealth managers. (To view the full report via a registration link and URL, click here.)

As previously discussed, wealth management advisors should really be looking forward to their work being facilitated by AI, rather than fearing replacement by it. However, it seems certain that AI - in combination with other new technologies - will indeed cut a swathe through compliance. 

Since 2011, US and European banks have been hit with $150 billion of litigation and conduct charges. As regulation has ratcheted up, the scramble to avoid censure has seen many institutions double their compliance headcount. Currently, financial institutions have to dedicate 10-15 per cent of staff to governance, risk management and compliance and regulation is now estimated to cost the industry an eyewatering $270 billion yearly and as much as $1 billion per firm.

Valiant efforts have been made with outsourcing and labour arbitrage, but we are only at the beginning of a huge wave of regulatory automation that will sweep the sector. “Industry observers agree that more compliance functions will become automated over the next 3-5 years as regulations themselves become more complex, multi-modal and extra-territorial in outlook,” said Dr Anthony Kirby, Associate Partner, Regulatory and Risk Management - Regulatory Intelligence at EY in the UK. 

The potential to simultaneously slash costs and business risks through regtech is increasingly becoming too compelling for institutions to ignore – particularly given the added chance to vastly improve the client experience.

Relieving onboarding pain
Much previous WealthBriefing research has focused on the acute challenges in the client discovery and take-on phase for investors, advisors and institutions. Onboarding is often inordinately lengthy and onerous due to a lack of digitalisation, leading 71 per cent of wealth managers to fear clients dropping out during the process.  

Client due diligence obligations have become increasingly weighty amid the global crackdown on financial crime and are the source of much of this pain (wealth managers have estimated that screening high-risk clients requires an average of 5.4 hours of work and even low-risk ones 1.6 hours). 

They have therefore become a big regtech focus. Some 63 per cent of wealth managers globally foresaw increased spend in this area in 2017, but with 44 per cent opting to concentrate on technology investment rather than throwing (ever-more expensive) personnel at the problem. And, with the advent of AI applications for Know Your Client (KYC) and Anti-Money Laundering (AML) purposes, wealth managers’ reliance on technology in these labour-intensive and high-risk areas seems certain to rapidly accelerate.

So, how can AI technologies help make client discovery and documenting due diligence both better and more efficient?

As our experts observed, the vast majority of prospective clients will present no real compliance issues that will call for escalation to a human expert, so there is great scope to automate the collation of all evidence required to onboard them. Documenting sources of wealth/funds constitutes a major headache for firms, while for clients the questioning required to open an account can feel intrusive and laborious. 

AI technologies can ensure they are only asked what is strictly necessary and facilitate an element of “self-service” (and depersonalisation) that might be very helpful in fledgling relationships. Importantly, 38 per cent of clients already prefer to open accounts digitally today, with this expected to rise to 52 per cent in the next few years.

As David Teten, Managing Partner of HOF Capital, argued: “Client onboarding is an extremely cumbersome and manual process that could be improved by new technologies in many ways. For example, implementing Robotic Process Automation can streamline KYC decision-making through more interactive and intuitive information-gathering. AI can also be a great help in mining public data sources to find out things like the value of a client’s home or of the company they sold.”

Although most clients are unproblematic from a compliance perspective, it must also be remembered that wealth management is by definition a highly cosmopolitan industry assisting clients with complex, international financial affairs. Here, AI can be invaluable in ensuring that all clients who can be onboarded, are, and that riskier ones do not slip through the compliance net.
Digital detective work

Just as in lead generation and news personalisation, one of the most powerful ways AI can be applied in a client due diligence context is in using Natural Language Processing (NLP) to “read” vast amounts of information in any language. “AI can be a great help in the onboarding phase, through intelligent document scanning and sifting through the array of external data sources wealth managers should be consulting,” said Alessandro Tonchia, Co-Founder of Finantix. “As well as massively reducing risk, it can hugely improve sales effectiveness and enhance the client experience.” 

As Tonchia explained, the real power of the technology lies in its ability to intelligently extract risk-relevant facts from a huge volume of data, but then to also synthesise and deduplicate that information so that it is both meaningful and concise. 

“Before, the technology might have flagged a hundred mentions of an individual doing business with North Korea, but now it will collapse those hundred documents into a single ‘red flag’ alert,” he said. “NLP can also discern the difference between a person having, say, starred in a film about terrorism and them having been actually linked to it. Eliminating false positives and irrelevant results makes analysing true risk a much easier task.”

In addition to summarising information, the massive reach of AI analyses can also uncover risk indicators that it would take an inordinate amount of detective work to uncover manually.

“One area where we’re really adding value is in network analysis,” Tonchia continued. “Here, a prospective client might present as totally clean, but we could discover that in fact they sit on the same board of directors as a very dubious individual, or that they are an advisor to a company that has a joint venture with a sanctioned entity, for example. 

“It’s about detecting third-level relationships and indirect risks to mitigate all conceivable risk factors. Criminals and sanctioned companies are, after all, unlikely to act in the open to try to open an account with you.”

The power of many of the AI applications explored in this report lies in Machine Learning (that is, where systems learn and improve from experience). But the sophistication of Finantix’s technology goes beyond even this in the ever-evolving fight against financial crime. 

“You need to not only define an individual’s network of relationships, but also to navigate and make inferences about the connections,” he said. “We’re now going beyond Machine Learning and investing heavily in reasoning tools and inference engines that emulate the ‘thinking’ of a human investigator.” 

Barriers beginning to come down 
The terrorist threat and a series of money laundering scandals are ensuring that the fight against financial crime remains absolutely top of the international agenda, while at the same time clients are growing increasingly intolerant of inelegant consumer experiences – no matter how good the regulatory rationale. Combined, these factors make compliance - and client due diligence in particular - among the areas most ripe for AI amelioration. And this must surely be right around the corner. 

Regtech solutions are still evolving, but, according to EY’s Dr Kirby, the main brake on adoption is that “the IT environments at regulators, central banks and governments are not yet at the point where they can readily interoperate with the industry as a whole”. However, he sees this rapidly changing as regtech solutions and “smart” or self-executing contracts become more mainstream (these are computer programming codes that facilitate or enforce the performance of an agreement using blockchain technology).

“Machine Learning, Artificial Intelligence and Natural Language Processing will quite soon be widely applied to the ‘second line of defence’ skills in legal, compliance and risk management,” he said. “This will start with day-to-day monitoring in activities such as surveillance and AML and extend over time to filing reports of trades and transactions to meet regulatory conduct of business and prudential obligations in each jurisdiction.”

Believing them beneficial to society as a whole, as well as to the institutions pooling their resources, regulators and governments are thought to be increasingly positive on the use of “utility technolo¬gies” (this is where service/technology providers offer a centralised outsourcing of key common tasks, potentially across the entire industry). In an AML context, this would mean institutions and authorities sharing KYC, transactional or other data through a third party “utility” - an approach which would likely have to draw on Distributed Ledger Technol¬ogy in turn. 

Compliance is undeniably the area most fraught with complexity when it comes to AI adoption, and great strides in other technologies such as biometrics and big data will also be required for real progress to be made. However, it is also arguably where wealth management sector stands to make the biggest gains. For many institutions, the compliance burden has become unbearably heavy and, as we have seen, AI has huge potential to help lighten the load.