Artificial Intelligence (AI) is emerging as a key technology for the North American wealth management industry. Alongside the rise of chatbots, AI is being deployed in digital advisory platforms (robo-advisors) and in systems assisting advisors in managing their clients. Robin Arnfield interviews key players in the AI space to get more insight into top use-cases.
It is no longer early days for Artificial Intelligence (AI) in wealth management. Indeed, what was an area to watch a few years ago is increasingly entering the mainstream, and most private banks and wealth managers are finding different use-cases for AI in their organisations.
The use-cases are varied. According to Rajesh Kamath, head of financial services solutions and incubation at US-based IT consultancy Incedo, AI allows firms to scale in a way that isn’t linearly related with their size, so they can do more with the same employee base.
“AI will affect wealth management firms’ front-, middle-, and back-offices. Using AI can help drive down a company’s total cost of operation in the mid- to long-term and improve advisor productivity by deploying chatbots as self-service channels,” says Kamath.
In North America, particularly, AI is already been deployed across multiple business lines. Here are the top proof-of-concept use cases, currently, for AI in wealth management.
A growing number of North American wealth management firms offer hybrid robo-advice platforms combining automation, AI, and human expertise.
These include BMO Bank of Montreal with SmartFolio; Charles Schwab with Intelligent Advisory; Merrill Lynch with Merrill Edge Guided Investing; and Wells Fargo Advisors with Intuitive Investor, which was developed in partnership with US FinTech SigFig.
“Intuitive Investor is currently in a small employee pilot, with an expanded customer pilot coming this summer, leading up to an official launch later in 2017,” a Wells Fargo spokesperson says.
A spokesperson from Charles Schwab says: “Schwab doesn’t use the term ‘robo-advisor’. We describe Intelligent Advisory as a hybrid advisory service combining the best of human and machine. It combines live credentialled professionals and AI to provide investors with a personalised financial plan, on-going guidance, and an automated and diversified portfolio.
“Intelligent Advisory is designed for mass-affluent investors without a complex personal financial situation. The target audience is digital-first, and is self-directed or wants to maintain control of their finances with the ability to engage with financial professionals when needed.”
Several FinTechs offer consumer-facing robo-advisory platforms including US-based Betterment and BlackRock subsidiary FutureAdvisor, and Canada’s Responsive.AI, Wealthsimple, and Nest Wealth.
As of April 2017, Wealthsimple had approximately 25,000 clients and C$750m ($557m) in AUM, a spokesperson says. “These figures include US clients, as we entered the US market in January 2017,” she says.
Randy Cass, Nest Wealth’s CEO, says: “Our technology satisfies the criteria for a robo-advisor as it’s a completely automated digital advisor platform. It handles everything from on-boarding to digitisation of paper documents to digital KYC and online portfolio allocation.”
Canadian wealth management firms Credential Financial and National Bank of Canada have agreed to deploy the Nest Wealth Pro platform.
“Nest Wealth Pro will be rolled out over the next few months,” a National Bank spokesperson says. “Our objective is two-fold: provide our advisors with an additional working tool and our clients with a better digital experience.”
US Bank Wealth Management is working with FutureAdvisor to offer a hybrid advisor platform, which is due to be rolled out this year.
Canada-based Responsive.AI, that white-labels its digital advisor and research technology for FIs, uses a model that is driven by “statistical learning” that examines patterns.
“By breaking the model into independent choices that make diverse decisions, using diverse data, we mitigate the potential for total model failure – which is what happens with the static co-linearity of passive investing,” says Davyde Wachell, CEO of Responsive.AI.
Arushi Srivastava, Senior Director, Digital and Cloud Services at systems integrator NTT Data adds: “Several banks that are our clients are piloting AI for back-office-enabling applications like tracking stock movements or correlating social media activity with stock movements.”
Chatbots and beyond
Chatbots use Natural Language Processing (NLP) to answer customer queries and action transaction requests, running on digital banking channels and messaging platforms such as Facebook Messenger.
Natural Language Generation (NLG), a technology generating natural-language narratives from computational inputs, is used to assist advisors in briefing clients. Vendors include Narrative Science and Yseop.
“NLG has great potential in wealth management as a time-saver,” says Celent analyst Kelley Byrnes. “Using NLG, advisors automatically generate natural-language fund or stock insights for clients that are derived from raw computational data, without doing analysis themselves.”
Cass, Nest Wealth, says AI’s role in digital advice is more on the NLP side than on portfolio construction. “AI’s initial application in wealth management is to provide an easier way for clients to have their questions answered and understand how products work.”
He adds: “We are currently working on NLP with IBM Watson, so we can offer a chatbot in Q4 2017. We’ll be able to feed the thousands of digital conversations we have with clients into this NLP solution.”
IBM Watson offers a cognitive agent – a chatbot that can be used to automate some aspects of advisors’ day-to-day tasks without calling their advisor, says IBM’s Senior Offering Associate – Financial Markets and Wealth Management, Alex Baghdijan.
AI can change the way advisors interact with clients and advisors and clients interact with wealth management providers, says Incedo’s Kamath.
“A wealth management client of ours is piloting advanced facial analytics to see if customers’ stated investment goals are really the goals they’re interested in. This will be used at the point of sale in goal-setting exercises between clients and advisors, as well as for investor education.
“In our Incubation Lab, we’re implementing a system that lets advisors and clients access their financial, product and other information via voice conversations with a chatbot. The bot uses AI to understand natural human speech, and accesses the appropriate information from the organisation to fulfill the client’s investment intentions,” he adds.
Eran Livneh, Personetics’ VP of Marketing says the company is focusing on using AI to provide “actionable insights and advice”, whether directly to customers or to advisors to assist clients.
“Our applications are about the customer engagement ranging from on-boarding and running customers through compliance questions to enabling them to select funds and move money between funds.
“Personetics has developed a cognitive banking framework, which offers not just a chatbot conversational capability but also an underlying analytics layer that connects to the bank’s transactional systems and understands who the client is, what they are trying to do and what information would be most useful to them right now.
“Our NLP and contextual analysis technology examines what the client is saying and also what their transaction data is to get an understanding of their entire financial situation,” explains Livneh.
Reporting activities and AML detection
AI plays to its strengths in the middle office where human discretion is in maximum use at wealth management organisations.
Some hedge funds acting as investment providers to the wealth industry already use AI techniques where portfolio tracking, rebalancing and investment selections/trade decisions are made in an automated fashion.
According to Kamath, one of Incedo’s clients is talking to an anti money laundering (AML) solution provider for its client on-boarding process, and the product uses AI to determine the probability of clients facing money-laundering risks using public domain data such as news items and court records.
Currently, reading and extracting actionable intelligence from documents require human action. However, using this information in buy-side processes can be offloaded to AI, with humans only required for complex cases.
“Additionally, fulfillment processes involving low human discretion, which are high-volume and repetitive, are already being disrupted by AI-based process automation. This scope will soon be extended to processes that require some human discretion.
“For example, complex data reconciliation between multiple statements from investment process participants like hedge funds and prime brokers is an area where AI can help in the near term. This is especially true in private banking, where investment vehicles can get really complex with multiple ‘regular’ and exotic asset classes and hierarchy structures,” says Kamath.
According to Responsive.AI’s Wachell, all banks are trying to work out how to “turn AI from a science project into profit for the bottom line”. Responsive.AI is currently developing high-resolution KYC technology using AI to detect personality traits a regular KYC wouldn’t spot, says Wachell.
“Our future focus is to use AI tools to get a deeper understanding of clients. Someone can state in a questionnaire that they make C$200,000 ($148,588) a year and seem stable. But, if you examine their bank account, you may see they have erratic spending habits. Having higher-resolution knowledge of clients would change the services you offer them.”
AI allows firms to gain new insights into their business, beyond traditional analytics. Kamath informs that Incedo is talking to a client to understand how public media content such as blogs frequented by advisors can be used to model advisor behaviour.
IBM has developed a “Client Insight for Wealth Management” solution that uses the cloud-based Watson AI system to process vast amounts of information and learn from each interaction.
“This is designed to augment advisors, changing the way they manage their day-to-day business and making them less reactive,” says IBM’s Baghdijan. “It lets advisors predict client attrition and know ahead of time that clients may be leaving so they can try to retain them.
“We have some Watson proofs-of-concept with advisors, and are moving forward into production with these clients,” says Baghdijan.
He adds there will continue to be a demand for human advisors but IBM’s role is to leverage technology so advisors can go from managing 100-200 clients to managing thousands while still offering personalised advice.
“In the future, as so many processes will be automated, advisors will be more like quarterbacks who have advanced tools and can call on experts to help clients in niche situations,” he adds.
IBM has also partnered with SalesForce to develop a Watson-based customer relationship management (CRM) solution for wealth management.
“We’re training Watson to provide advisors with a unified dashboard so they can manage their clients. Typically, advisors have client data in many different systems, so our answer is to put all this in a single dashboard.
“We also offer a segmentation tool so advisors can segment their clients based on their different financial behaviour attributes, such as people who typically call when the market falls,” says Baghdijan.
Challenges to overcome
Every wealth management firm has different challenges and opportunities around where AI can lead to maximum benefits.
“It needs someone with a vision of how AI can help to analyse organsiational factors and build an AI strategy. In our experience, this is currently lacking in most firms,” says Kamath.
He explains that a near-term cost spike can be expected during the wealth management firm’s AI implementation, as “there’s typically a stabilisation period for AI”, where the deployment is iteratively refined in a sandbox before being brought into full operation.
“In this phase, the organisation incurs both AI and human (operations team) costs. Both costs are likely to fall after this phase, and reach a lower steady state.
“Also, obtaining good AI skill sets in programmers is difficult and expensive,” adds Kamath.