What if artificial intelligence could tell you that your client was exhibiting some of the behavioral patterns of other advisory clients who have fired their advisors? Actually, this capability already exists, and it’s one of the ways that AI will change the RIA business.
Many, if not most or all of the platforms serving financial advisors offer A.I.-generated tools. BNY Mellon | Pershing’s platform, for example, offers A.I.-assisted account transfers designed to make that process much easier. Another feature on that platform involves trade surveillance where A.I. alerts an advisor about a trade they’re trying to make that might not pass the sniff test with compliance. One particularly intriguing A.I. tool sends signals to an advisor that a client is not happy with the relationship.
As to how A.I.-generated data interacts with advisors, Ram Nagappan, chief information officer at BNY Mellon | Pershing, says when an advisor looks at a client’s name on their dashboard they’ll see various color-coded signals on the side representing a prioritized list of information, or indications about the client derived from the A.I.
As for those signals pertaining to the client relationship, an assessment is delivered to the dashboard in the form of a relationship health score that translates risk (i.e., client unhappiness) into a range ranked from highest to lowest. Some of the factors used to assess relationship risk include recent account transfer requests to other financial institutions; the performance of the client’s accounts relative to the overall market; and poor responses to client surveys and feedback, as well as chat sentiments.
Other factors include how frequently an advisor met with the client during the past year, and whether a financial plan exists and, if so, when it was last updated.
In addition to these factors, Nagappan explains, A.I. can analyze historical examples of clients who already left their advisors and can look for additional signals that may be relevant to assess client relationship health.
And so artificial intelligence, which seems so futuristic, is very much in the here and now in the advisory profession. But what exactly is it?
Demystifying A.I.
During a two-hour broadcast on PBS’s investigative documentary series Frontline that examined the perils and promise of artificial intelligence, the show’s longtime narrator, Will Lyman, whose sonorous voice would likely lend an aura of drama and gravitas even to his reading of a dinner menu, had this to say about the subject: “The A.I. algorithms are ushering in a new age of great potential and prosperity, but an age that will deepen inequality, challenge democracy and divide the world into two A.I. superpowers.” Those being the U.S. and China.
That’s ominous stuff. As A.I. has continued to amaze—if not alarm—the world since the Frontline episode aired in 2019, the technology is increasingly gaining traction in the financial services world. And that includes the wealth management space, as the Pershing examples show. That said, many people within the industry still don’t know what to make of A.I., or even really know what it is.
“When I meet with CEOs of [wealth management] organizations, I ask if they can tell me what A.I. means, and they shrug their shoulders and go, ‘I don’t know, but I read about it all the time and I think I have to have it,’” says Brandon Rembe, chief product officer at Envestnet, whose wealth management platform offers advisors a variety of A.I.-driven tools.
“It used to be ‘big data’ was the buzzword,” he says, “but it has shifted toward A.I. A lot of people conflate A.I. with business intelligence reporting, but they’re very different. I think the industry is trying to figure out what A.I. is; it’s a hard definition to pin down.”
A.I. is a broad term that’s classified into branches, subsets or types. And the number of these different categories can vary, depending on the source. According to IBM, any system capable of simulating human intelligence and thought processes is said to have artificial intelligence.
Machine learning, which is a branch of A.I., employs algorithms enabling computers to parse data, learn things and perform human-like thinking—presumably at a level that exceeds human capability. For example, the IBM supercomputer named Watson used machine learning to defeat the two all-time champions of the game show Jeopardy in a two-round match in 2011.
Other A.I. branches include deep learning (a more advanced form of machine learning using neural networks applied to larger data sets), natural language processing (which helps machines understand human language), fuzzy logic (a reasoning method that converts subjective, or “fuzzy” input into concrete results) and robotics, among others.
According to people in the know, A.I. requires a feedstock comprising reams and reams of data—coupled with smart individuals to turn that data into algorithms—to create actionable insights and real-world solutions.
“At Envestnet we have millions of investor accounts and analyze billions of transactions every month and have multiple trillions in assets that we look at,” Rembe says. “Those are the types of numbers you need to look at to scale A.I. and have it be meaningful.”
Envestnet uses A.I. in several ways including collating and analyzing various financial activities across an investor’s entire financial picture beyond just investments, and then employing what Rembe calls hyper-personalization, or tailoring specific action for a certain client or client type based on the data gleaned about their spending habits, credit risk levels, personal interests, investment preference or other attributes that enable an advisor to provide targeted service.
“That helps the advisor suggest the next best action they should take for a client,” Rembe says.
Envestnet also looks at how wealth management firms interact with clients via A.I. One of its offerings to advisors is a chatbot, added in 2019 when it bought Abe.ai and its A.I. assistant with a voice-and-message enabled interface. Chatbots use software that employ natural language processing and machine learning to communicate with people via voice or text.
“We’ve seen big adoption among wealth and financial institutions, and even in retail banking, where A.I. and the conversational interface has gotten good enough where clients expect to be able to not just click through a bunch of buttons but to have a conversation with their technology and get more answers in an omnichannel environment,” Rembe says. “Things like this are the next generation of where people are going. And people are getting more comfortable with that and expecting that conversational user interface.”
Hype, Or Reality?
Sounds great, but have wealth management firms bought in to the A.I. concept? The picture is muddy, at least according to a report put out late last year by Accenture that surveyed 100 technology and business C-level executives at wealth management firms in the U.S. and Canada.
Among the report’s nuggets: 84% of respondents believe A.I. will transform their industry in the next five years, but just 28% are currently scaling it across their businesses, and 85% believe the promise of A.I. within wealth management is currently more hype than reality.
According to the survey, the technology is seen more as something to use behind the scenes than for applications that face clients. Specifically, 65% of respondents believe A.I. can create the most long-term value in the back office, while only 35% see it as a way to boost the client experience and engagement.
“They’re using A.I. effectively in narrow-sleeve use cases,” says Scott Reddel, managing director and leader of Accenture’s wealth management and capital markets practice. “The maximum value being reported is the middle or back office relating to automation, efficiencies and those kind of things. It’s definitely more tangible there, because you can see where the efficiencies are being created and where the cost reductions are.”
Indeed, Accenture estimates that around 30% of an advisor’s daily activities could be automated through A.I.
“But no one is really using it at scale across the breadth of the value chain or rolling it out to their advisors in a meaningful way,” Reddel says. “Our view is that A.I. will do for the next decade what digital did for wealth management during the past decade in terms of transforming the advice that has been delivered. We did some of our first A.I. work with wealth managers four years ago, and there have been incremental improvements rather than a wave of change.”
Reddel notes that big wirehouses, large broker-dealers and the B-Ds attached to large financial services firms are focused on building A.I. capabilities themselves. They have the resources to hire the people or acquire the technology to make it happen, though it has been slow going thus far.
“A.I. has been on the agenda of some of the bigger players for the past few years, but there has been a bit of organization fatigue in terms of when they’re going to realize the value from that,” Reddel says. “You have to invest a material amount, and there’s a bit of lead time required to create the foundation that unlocks some of that value.”
Wealth management firms in the mid-market and independent space, he adds, can tap into companies and platforms that provide A.I.-related insights and tools that can be tailored to their advisors’ needs.
One RIA’s Solution
Necessity being the mother of invention, some financial advisors have taken the initiative and shown that they don’t have to be Fortune 500 companies or Silicon Valley upstarts to develop effective A.I. platforms on their own.
Andrew Altfest, president of Altfest Personal Wealth Management in New York City, worked with technologists and more than 40 subject matter experts to develop FP Alpha, billed as a wealth management solution that enables advisors to scale their holistic planning efforts across 17 financial disciplines in four main subject areas: planning (including estate, divorce, cross-border planning, etc.); savings (including expense analysis); protection (which involves various insurance products); and lending (which includes student debt and mortgages).
The idea for FP Alpha came to Altfest while he was trying to do holistic planning for clients beyond retirement planning and investments and reached a “pain point.” It was a massive time crunch working on taxes, estates, insurance, lending and the like, which all require checklists and spreadsheets, as well as the time to read wills, trusts, insurance policies and tax returns.
“To apply holistic planning, you have to do this work manually,” Altfest says. “There’s no way to scale it, and perhaps you have time to do this type of planning only for your wealthiest clients. I didn’t understand why software wasn’t serving this space. I connected with other financial advisors, and they basically said, ‘If you build it, we’ll use it.’”
So he consulted with subject matter experts in various financial fields and hired tech people who use machine learning and natural language processing to convert the experts’ insights into algorithms.
“When you enter your client data into the software, you have an integrated wealth management team of attorneys, accountants, and insurance and mortgage brokers who are looking into your client’s situation because their brains are in the software through the algorithms,” Altfest explains.
It works like this: An advisor uploads client financial documents such as wills, trusts, tax returns and insurance policies. FP Alpha uses A.I. to extract information from those documents and then formulates recommendations to advisors, telling them where the planning opportunities are.
For now, the program costs advisors $1,500 for the first license and $1,000 for additional licenses. Altfest says those are introductory rates, and that the list price is $1,000 more for both pricing structures. Advisors license the program directly through Altfest’s firm. He says he began building the FP Alpha software in 2017, has been using it at his firm since 2018 and launched it publicly in February 2020 at the T3 conference. Altfest says FP Alpha’s development costs were “easily into seven figures.”
The product has received very positive advisor feedback, he says, adding that enhancements are coming later this year. And while he doesn’t disagree with the notion that A.I. needs significant data to produce meaningful and actionable results, he believes many RIAs generate enough digital data from their businesses to develop their own A.I. capabilities. (Altfest Personal Wealth Management is a sizable firm with a lot of numbers to work with: It manages about $1.5 billion from around 3,000 accounts, along with roughly $15 million from about 85 accounts on a non-discretionary basis.)
“You need data, but it depends on what you’re trying to accomplish,” Altfest says. “From there you can figure out how much data you’ll need.”
Dr. Strangelove
Despite the good things A.I. seemingly can do for financial advisors and the world at large, it nonetheless creates an uneasy feeling in some people. After all, artificial intelligence conjures images of science fiction run amok where algorithms (and governments that use these algorithms in a heavy-handed way) control our lives and compromise our privacy.
“There’s something in A.I. called the ‘creepy factor.’ It’s a thing,” says Rembe from Envestnet. “People ask, ‘How do you know all of this?’
“One thing we spend a lot of time on is our A.I. governance model,” he continues. “If you just have A.I. without human oversight and the governance model, you can run into things like gender and racial bias that can negatively impact consumers. You can’t just let A.I. run wild because it’s built by humans and can make mistakes like humans do. And unless you have a human to look for and evaluate those mistakes, it will make those mistakes.”
Ram Nagappan at BNY Mellon | Pershing says his firm puts an emphasis on removing the mystery of how A.I.-driven results are derived.
“With the A.I. models that existed, nobody could explain how it came up with its results. It was like a black box,” he says. “Now, people are working on providing what I call an ‘explainable A.I.’ that gives people the confidence to adopt these things. If you don’t understand it, sometimes it might not work right and you might make wrong decisions. Compliance teams want to know how decisions are made.”
On both the Pershing and Envestnet platforms (and presumably other platforms offering A.I. solutions), some of the tools are baked into the overall package that advisors sign up for, while other solutions cost extra. Regardless of how A.I. is packaged and used, it appears that advisor-centric platforms are engaged in an arms race involving artificial intelligence.
“A.I. will continue to be a competitive differentiator for what Pershing has to offer versus Schwab or Fidelity,” says Reddel from Accenture. “They have been competing on their digital platform and access to products. I think A.I. insights will be the next frontier on how to compete.”
A lot of people are gung ho about A.I.’s potential benefits for the wealth management profession. After all, it’s part of the inevitable progression of technology in this industry. But you can’t fault people for being wary of getting too cozy with this potentially all-consuming technology. One of the takeaways from the Frontline investigation—and something that has been discussed often in many other circles—is the likelihood that A.I.-driven automation will destroy countless jobs across many industries. While it seems unlikely that robots and algos will someday displace the friendly neighbored financial planner, who would have thought robots would be viewed as a solution for eldercare? (And can A.I.-linked algos create Ponzi schemes that make Bernie Madoff look like an amateur?) For now, these are probably just hyperbolic ruminations.
“We believe there will always be a place for human-led advice,” Reddel says. “The way advice will manifest itself in the future will be materially different, and without A.I. it will be hard for most advisors to compete in terms of providing the right insights.”
Perhaps someone should create a movie called, “Dr. Strangelove or: How I Learned to Stop Worrying and Love A.I.”