Hal. Skynet. The Matrix. Science fiction is replete with tales of artificial intelligence run amok.
However, as a new generation of AI-backed tools proliferate in the wealth technology space, advisors are more likely to find that these tools are time-saving aids, not a threat to their existence.
But we should also demystify this technology. For one thing, most of what’s called artificial intelligence in 2023 isn’t particularly intelligent. For another, there is no single technology that deserves the name.
“It’s really a pool of AI techs,” says Ritik Malhotra, co-founder and CEO of Savvy Wealth, a digital platform for financial advisors. “People conflate it with the idea of AI as if there’s just this one technology, like the cloud, but AI has many different variants. The most interesting to us is the latest to come into awareness. It’s what people call ‘generative AI’ or, more technical, large language models.”
In the wealth management space, the technology has been around awhile. AI-related technologies such as natural language processing and machine learning have been used to automatically analyze client documents or quickly surface recommendations for the next best actions by financial advisors. RiXtrema, an analytics platform for fiduciary advisors, has been using legacy AI applications like machine learning for more than eight years via its 401(k) fiduciary optimizer—a platform that analyzes workplace defined contribution plan fund menus and lineups to find places where lower-cost funds or more appropriate ones should be swapped in.
But generative AI is a newer and even more impressive technology that’s freely and openly available to nearly any computer user. It includes programs, the most famous being ChatGPT, that not only read and analyze data written in human vernacular but also synthesize content in natural language or images based on that information, says Yon Perullo, RiXtrema’s chief executive officer. Today, ChatGPT can be freely accessed via a number of channels, with the easiest likely being Microsoft’s Bing search engine.
Some, if not most, advisors are already using this new technology, according to survey results released in June by SmartAsset: Out of 100 advisors surveyed, 27% said they were currently using ChatGPT or other generative AI products. Of these, 44% said they were using the technology to write educational blog content, 28% were using it to follow up on leads with prospects, and 28% were using it for routine communications and scheduling.
Perullo says, “Once something like ChatGPT comes out, where you can ask it all these questions, we began to ask how we could leverage something like that, that doesn’t know about plans or portfolios, in a way to make our clients’ lives easier. When we say we created a product around AI, it’s a bit of a misnomer because all we’ve really done is teach a machine the relationships between words—there’s no actual intelligence.”
RiXtrema is using AI to help advisors analyze 401(k) plans and list out the advantages and disadvantages of the plans’ designs, then it automatically drafts emails to pitch better alternatives to plan sponsors. RiXtrema’s applications can also write call-in pitches for advisors and explain compliance rules they may not be able to translate into layman’s terms for prospects and clients.
How It’s Used
Companies such as Savvy Wealth and Envestnet have found three main uses for generative AI: in marketing, client relationship management and compliance.
“The first way we saw advisors using generative AI is to create content—or at least create the idea for content,” says Dani Fava, the group president of product innovation at Envestnet. “You can go to something like ChatGPT and ask it to act like a financial advisor who has high-net-worth clients who are thinking about summer vacation. It will tell you what it thinks your clients most want to know about summer vacations—like, are they most interested in budgeting or whether or not they should buy a summer home. Then you can turn around and ask ChatGPT to write a 500-word newsletter about the benefits of renting versus buying a summer home, take that edit and send it to customers.”
Generative AI can also be used to create blog posts, imagery and other content. According to Scott Lamont, head of consulting at RIA wealth tech consultant F2 Strategy, software using generative AI combined with machine learning can both draft emails and pitches to prospective clients and also sift through a large number of potential prospects to find the ones that best match an advisor’s most successful clients.
Malhotra says Savvy Wealth has used AI for email responses. “On working with prospective or existing clients, we’ve launched functionality recently where if clients email their advisor, we’ve built in an integration in our technology using generative AI to create email draft responses back to the client, given all of the data, all the past actions, and the potential responses,” Malhotra says. After the AI generates a response, advisors can edit it, putting it in an advisor’s voice and further customizing using their personal knowledge and intuition about the clients.
F2 is also exploring the ways AI can be used to sift through market data to create a general assessment, and then align that assessment to clients’ data such as their portfolio holdings, Lamont says. An advisor can then use generative AI to create talking points for the next client call or meeting, all in about five minutes.
Similarly, in compliance applications, generative AI’s ability to understand and summarize large amounts of content can help create more efficient processes. “AI can also be trained to look for patterns of behavior and to identify potential bad actors,” says Adrian Johnstone, CEO of Practifi, a business management platform for RIAs.
However, he warns that the technology also gives these bad actors more weapons as well.
“Just as AI can bring the good guys scale to help create better compliance and cybersecurity, it also gives the bad guys scale to exploit vulnerabilities,” Johnstone says. “Generative AI could make things like phishing campaigns much more mature and effective. It can be used to power ransomware attacks and those types of things, too.”
RiXtrema is also using AI to monitor email for compliance red flags, says Perullo. And he warns advisors to be careful on bots like ChatGPT, since any information they give it will be absorbed and made publicly available. That means public AI models are not good places for anything involving private or protected information. RiXtrema has built its own model, and thus hopes to enable advisors to deploy AI more directly in their client relationships and the portfolio management experience, taking the client’s actual demographic and financial data, in real time.
But AI is not a tool that can “do it all” in financial planning and wealth management, warns Johnstone.
“Generative AI and natural language processing are not well-suited to analyzing huge volumes of numeric information,” he says. “A lot of people think when they hear about advice tools and AI that they’re going to get these deep market predictions and analyses of huge sets of numerical data, and it’s just not designed to do that. It’s more about drafting emails and blogs and content creation, and analyzing language-based inputs.”
What’s Next
Malhotra emphasizes that the wealth management industry is still just figuring out how to deploy generative AI, which he describes as “Phase 1.” Phase 2 will come when AI is used to produce something entirely new, something that may currently be well outside of the realm of financial advice or wealth management experiences, that augments or alters the client or advisor experience.
Of course, it’s hard to say exactly what that “Phase 2” experience or product might be, but he argues that it could be a more concierge-like experience where an advisor, by virtue of the analytic power of AI and the efficiencies it creates, is able to take on the management of a client’s household in ways that go beyond addressing their financial needs.
Similarly, most of the technologists Financial Advisor talked to who are working with AI believe that it will augment, not displace, knowledge workers like financial advisors. With the help of generative AI, content will be created more efficiently, with greater ease and (eventually) with greater accuracy—but any content generated by AI will still need to be carefully edited.
That is due, in part, to generative AI’s tendency to “hallucinate” or fabricate data that looks and sounds real but was really created out of thin air. While such behavior can be reduced or curtailed by carefully establishing parameters around AI programs, there is no foolproof solution.
“Where is generative AI most useful? When you have a task that is somewhat deterministic, that has a deterministic output,” Malhotra says. “So something like writing software code, you write code, run it and you get output, it’s very deterministic, and AI is excellent at making that way faster.” However, he adds, “At the end of the day, there’s a lot of subjectivity in financial services that AI is not going to be able to displace.”
In the near term, that means the tech will simply make human advisors more efficient—and that will help them solve a demographic crunch facing the industry, as the number of advisors declines even as the demand for their advice grows.
“This is something that is going to help advisors scale and ultimately serve more customers,” Fava says. “Generative AI could be a tipping point—a piece of technology that finally really does free advisors to meet face-to-face with more customers—but it won’t replace advisors. It will be a force multiplier.”
Yet, at the same time, AI is likely to spur a new acceleration in the development of wealth technology advisors can use to serve their customers and streamline their businesses, Malhotra says.
“Our take is that it will dramatically speed up the development of any wealth tech,” he says. “We can produce new technology and launch new products weekly with advisors and clients, and we can do more with smaller teams. There will be more acceleration in wealth tech products.”