By Steve Garmhausen

Technology, from cloud computing to virtual meeting platforms to automated portfolio management, has helped smaller wealth management firms to compete with the industry's biggest players. But generative AI seems poised to give large businesses an advantage, at least for now, says Bradley Kellum, the head of wealth management for North America at global consulting firm Oliver Wyman.

"I do think the larger firms at this stage of the curve are advantaged, because they have the resources to test and learn," he says. "There clearly is value in centralized efforts toward digital marketing, lead gen, managing agencies, working in different geographies, applying best practices."

Speaking with Barron's Advisor, Kellum, who was a Charles Schwab & Co. executive before becoming a consultant, explains why we shouldn't expect gee-whiz uses of AI in wealth management. He discusses ways that leading firms are using generative AI, and which human roles it's most likely to displace. Finally, he offers advice on how wealth management leaders can cut through the hype around AI to decide whether and how to use it.

Where does the wealth management industry stand as far as generative-AI adoption? It's fair to say that we're still very much in the early days. Wealth will move more slowly than many other industries. Because the root of everything [in wealth management] is trust. If AI breaches that trust, the client will be the ultimate judge and jury. Wealth managers will be very protective of their client relationships and will demand oversight of technology. When you add regulatory considerations on top of this, there is a lot to unpack.

What are the most common generative AI uses at this point? Gen AI provides the ability to search for information more efficiently, a better user interface and user experience, the ability to interact with humans, to interpret information and learn patterns, to create custom content, and to summarize and organize information. If we think about those key capability areas, the part of the value chain that we see cutting-edge firms adopting most is in the service of supporting advisors.

Firms are having AI listen in on client conversations, take notes, figure out next best actions and follow-up items, and put things on the calendar. A lot of firms are implementing Microsoft Copilot for emails, or for synthesizing incoming information to make it easier to consume. Some are using AI in creating a first draft of client reports.

Some of the larger firms that publish thought leadership are using gen AI as a way of accessing and summarizing their points of view. So advisors can access that firm view to stay on script -- what is our take on tariffs, what is our take on the interest rate outlook, that kind of thing.

Massive productivity enhancements may be possible with all types of AI, and what this potentially unlocks is an advisor and client service workforce able to spend more time with clients, as well as develop a pipeline of new clients.

Sharing insights rapidly to the advisor is one level of how gen AI will help asset managers perform better. But using AI output to directly communicate this insight to the client relative to their unique situation is something that is not available to advisors just yet. Lack of proper supervision of content and recommendations generated by AI is a huge area of risk and still requires significant human intervention.

So is gen AI already making a big impact within the industry? Overall, people believe that current use of generative AI in the wealth management industry is more widespread and potent than it actually is. There are certainly gen-AI tools that take notes in client conversations, but we are just starting to see firms using AI agents to, not just transcribe, but synthesize that information and figure out relevant next steps.

As I mentioned, gen AI can also be used to create reports, but it's one thing to create a report for internal consumption versus a client-facing report. Creating reports for a client requires additional compliance review. When we discuss gen AI, the industry is understandably focused on exciting new possibilities, but we cannot lose sight that new capabilities need to be incorporated into the broader operating model, and that includes risk, controls and compliance frameworks.

What's the most sophisticated use of AI that you've seen implemented so far? It's where larger firms are marrying multiple technologies together -- generative AI in combination with machine learning. They might, for example, look at their library of client profiles for people who look similar [in terms of their financial picture]. The idea is to find ideas to bring to a client so the advisor can say, "I've noticed that some of your peers are doing X or Y." Identifying outbound opportunities, if you will.

What are examples of other advanced uses? The most advanced solutions involve the mirroring of different technologies, and mapping of language not just to a task, but to a workflow. One example is the transfer of accounts. The client can drop a statement from their current brokerage account into a secure Dropbox account. The software reads it and then prepopulates a transfer account form, which they can do in the same chat. It is an instance in which automation -- advanced optical character recognition and natural language processing -- and gen AI are being used to accelerate and streamline a workflow.

We are also starting to see new ways for firms to share hyper-personalized recommendations fueled by AI. New tools are emerging that take a view of a client's portfolio and provide insights using AI. In this scenario, the underlying data has been vetted and is compliance-approved. These tools are currently available for individual investors and will soon be available for advisories. They provide new ways to communicate with clients in a personalized, compliant way that is also cost effective.

Do larger firms have an advantage in the gen-AI race? I do think the larger firms at this stage of the curve are advantaged, because they have the resources to test and learn. And while many are using off-the-shelf large language models, they are training them on their own information. There clearly is value in centralized efforts toward digital marketing, lead gen, managing agencies, working in different geographies, applying best practices.

Some of these things lend themselves more to firms that have scale. And the platforms that serve the smaller firms will look to provide some of those capabilities, using their scale to support the investment in those technologies. The challenge is that firms need to have done the foundational work around data management, data quality and AI governance. Even the largest firms are early in their technology journeys.

Let's say I lead a large advisory firm and I want to start tapping into AI solutions. How difficult should I expect implementation to be? Large firms may go to Microsoft or OpenAI, for example, and buy an enterprise license. Depending on your goal, it can be a significant undertaking. You need a dedicated team of technologists. And it's not like you set it up and forget it. AI solutions are continuing to evolve and learn with continuous prompt improvement, and the models themselves continue to get better. As they do, there's a cost as a consideration as to which model you're going to use and how expensive it is based on what you're using it for.

Nobody I talk to thinks that AI will replace human advisors. But how much of their work do you think it will eventually absorb? It's funny, before I worked here, I worked for Schwab for many years. There was a time when our capital markets area was about 300 people and a couple of computers, and it turned into 300 computers and a couple of people. No one thought that the computers could automate block trading, yet they did.

So I am firmly of the view that technology will disrupt where it can, but that humans will continue to be vitally important. I think AI will get better and become more widely adopted in client service and improving operational effectiveness. The assistant role and perhaps the operational roles are most at risk of being outright disintermediated. But again, the value of the advisor is in that trusted relationship. There are a lot of different workflows that add up to that. And we've seen that when AI gets adopted, it's use case by use case, workflow by workflow. So over time you should see advisors get better at what they do thanks to AI. But that role of the trusted human will still be vital.

I could also see a world where you could click on an AI training video of what looks like a person, who is answering questions or showing you how to do stuff. You'd have to let clients know it's AI generated. You don't want to try and pass something off as real when it's not, because then clients won't trust you at all, and you'll be done. You could see somebody try the model of directly letting AI manage money. In some respects we've already seen a version of that with pure robo advice. A lot of people are sitting in retirement plans, and not all of them have enough assets to be interesting candidates for a traditional advisory relationship. Does AI create opportunity for a new service model for that profile?

Is the industry adopting AI slowly enough to guard against risks to client privacy? My observation has been that in general the larger firms, where more of the assets sit, are moving carefully. It might be that smaller firms flying under the radar take more chances. No doubt some will try.

What are some of the gee-whiz use cases for AI that might lie ahead? Consultants are often interested in the shiny object. But I think the best use cases are probably more boring. The difference for wealth management is generally not in the alpha of what you're providing a client. It's in trust and consistency and showing up.

(MORE TO FOLLOW) Dow Jones Newswires

June 13, 2025 11:13 ET (15:13 GMT)

Copyright (c) 2025 Dow Jones & Company, Inc.