At Future Proof, we heard a great deal about what AI can do for advisors. A story from The Real Adviser Podcast asks what happens when clients use it to examine the advice they receive.
By Alex Sullivan | Modern American Advisor
Last week at Future Proof in Huntington Beach, it was almost impossible to have a conversation about the future of financial advice without talking about AI.
We heard how it could pull information from a firm’s technology stack, prepare advisors for meetings, spot a client who needs a call and take some of the administration out of serving them. I sat in the audience for Anthropic’s “End of the Swivel” discussion with Schwab and Dynasty. The promise was compelling: give advisors better information and more time to use it well. We also covered Claude for Financial Advisors, including its proposed workflows across CRM, planning, portfolio and client communications systems.
Then, back from California, I read a story that made me laugh and stopped me in my tracks.
It came from episode 106 of TRAP: The Real Adviser Podcast. Andy Hart told his fellow hosts about an advisor who received a lengthy email ahead of a client meeting. It ran through retirement spending, cash flow, pensions, fees and investment questions. The client had apparently put the advice into an AI tool and sent the resulting list to the advisor. The hosts gave this rather unwelcome arrival a name: an “AI slop grenade.”
You can picture the email. It is late in the day. You have a meeting coming up. The client who normally wants to talk about when they can retire is suddenly asking for a technical explanation of the yield curve.
It is funny. It is also an excellent way of turning last week’s AI discussion around.
We have spent so much time asking what happens when advisors use AI. What happens when clients use it? What if the next application in your advice process belongs to the person across the table, and you have no idea what they entered into it?
This is already a live question. In “Who Do Clients Trust Now? AI, Advisors, and Expert Systems,” Dr. Sarah McLaughlin examined how clients now weigh a human advisor’s expertise against answers produced by intelligent systems. The TRAP story puts that question into an ordinary, slightly awkward client meeting. The client has brought a second voice into the room. The advisor has to decide how to respond.
I think there are four things worth taking seriously here.
1. Find the question behind the questions
The first reaction to a fourteen-point AI generated email may be irritation. I understand it. An advisor has spent years learning a client’s circumstances and building a plan. A chatbot has seen a fraction of that information and produced a list that seems to demand a written defense of everything.
The TRAP hosts did not agree on how far an advisor should engage. Nick Lincoln argued for a firm boundary around questions that the client does not understand or care about. Carl Widger pointed out that people already use AI to check quotes and proposals in other parts of their lives. His practical question was what the client hoped to learn from the list. Andy described the advisor’s eventual reply as brief and calm: he was looking forward to the meeting.
That last approach struck me as sensible. An email does not have to become a fourteen-part email exchange.
Imagine the client comes in and the advisor asks, “What made you run the plan through AI?” There are several possible answers. Perhaps they want reassurance that they can afford to retire. Perhaps a market fall frightened them. Perhaps their daughter has asked why they pay an advisory fee. Perhaps they were simply curious.
Each answer leads to a different conversation. A lengthy response about the bond market will be wasted if the real concern is whether the client can keep helping a parent with care costs.
There is a danger in treating the AI list as a personal insult. A client who could not have written a technical question themselves may still have a legitimate concern. AI has given them vocabulary, although it may have supplied far too much of it. It can make a worried person sound like a hostile examiner when they are really asking, “Am I going to be all right?”
I would start there. Ask which two or three points matter to them. Ask what decision they are trying to make. If the list raises a genuine issue that needs analysis, take it away and do the work. If a question rests on a false assumption, explain that without making the client feel foolish for asking.
An advisor does not have to follow an automatically generated agenda from top to bottom. They do have to listen carefully enough to find the human reason it was sent.
2. Make the advice easy to examine
The comedy of the TRAP example should not obscure one uncomfortable point: some AI generated questions will be good questions.
Hart described requests for detailed cash flow information and a complete account of charges. The podcast discussion moved from the nuisance of the email to a more serious distinction between firms doing substantial financial planning and those relying chiefly on investment selection. If a client asks why a recommendation fits their life and what they pay in total, the advisor should be able to answer.
That is particularly relevant to RIAs. We talk a great deal about the value of advice, but the value has to be visible in the work. A client ought to be able to see the purpose of a recommendation, the assumptions behind it, the alternatives considered and how it relates to the decisions they face.
Take a retirement income plan. A projection may show a favorable probability of success. That number is useful, but it is not a promise. The more valuable conversation concerns the choices underneath it: how spending might change, what happens after a poor run of returns, which expenses are flexible and when the plan should be revisited. If AI gives the client a different percentage, the advisor can examine which assumptions produced it.
Or take fees. An AI tool may compare the cost of an advised portfolio with a low-cost index fund and announce an obvious saving. The comparison may miss planning, tax coordination, behavioral coaching and work with the family. It may also expose a fee arrangement that the firm has explained poorly. Both possibilities deserve an honest look. “The software doesn’t understand our value” is a weak answer unless the advisor can show what that value is for this client.
Our Future Proof coverage of Anthropic’s panel considered what happens when AI draws information from across a firm’s systems. One point stayed with me: when connected records conflict, the system should surface the disagreement for an advisor to resolve. That is good discipline for client advice too. A second reading of a plan may reveal an outdated beneficiary, a changed family circumstance or an assumption nobody has revisited for years. The response is to check it.
AI will get plenty wrong. Sometimes it will identify something the firm should have noticed. The quality of the planning process determines whether that discovery feels like an ambush or a useful part of looking after the client.
3. Recognize the new person in the room
Sarah McLaughlin’s article gives this story a wider frame. She writes about advisors as an access point to complex expertise. Clients have relied on a person to explain financial decisions they cannot reasonably master alone. AI gives them another way into that knowledge, with all the possibilities and problems that follow.
That does not mean a chatbot has become the client’s advisor. It means the client can arrive at the meeting with an interpretation already formed.
The AI answer may sound definitive because it has converted uncertainty into a neat paragraph. It may have missed a crucial fact. It may have misunderstood the account, the household or the question itself. Even so, the client has read it. If it says the fees look high or the retirement plan looks risky, that impression becomes part of the conversation.
I saw a different side of this at Future Proof. The most interesting presentations were concerned with getting the right context in front of an advisor: the plan, CRM notes, portfolio information and events in the client’s life. Our report on Claude’s proposed advisor workflows described a system that could assemble those pieces for meeting preparation and propose follow-up actions after the conversation. Its usefulness depends heavily on what information the firm can connect and trust.
Now think about the client using a general AI tool at home. It is likely to have much less context, and the advisor may not know what was shared. That is the point to explore together, without turning the meeting into an argument about technology.
“What information did you give it?”
“Did it know about the income you need from this account?”
“Did it know why we agreed to keep that money accessible?”
“Which part of its answer concerns you?”
Those are fair questions. They allow the advisor to assess the output while showing respect for the client’s initiative.
Family dynamics make this harder. An adult child might review an elderly parent’s statements, run them through AI and come back convinced that the advisor has overcharged or overlooked an opportunity. The child may be mistaken. They may also have spotted a legitimate issue. Either way, the advisor has to work within the client’s wishes and permissions while taking the concern seriously. A generic answer from a machine cannot settle what the parent wanted or what was discussed over years of planning.
There is a privacy dimension too. Joe Spiers wrote for us about unapproved AI use inside advice firms and the risk of sensitive information entering tools outside a firm’s oversight. Clients may be doing something similar with their own statements and plans. Firms should be able to discuss what information a client chooses to share, in plain language and without scolding them for using a tool that has become part of everyday life.
4. Prepare for the meeting you are now going to have
An advisor may receive an AI generated list tomorrow. The useful preparation is quite ordinary.
First, make the plan understandable to its owner. Can the client find the main decisions, the reasons for them and the assumptions that would cause the firm to change course? Can they see what services they receive and what they pay? A document that satisfies the firm’s process but leaves its client unable to explain the plan invites someone else to explain it for them.
Second, create space in reviews for outside questions. Ask what the client has read or heard since the last meeting. Invite them to bring in a claim that worries them, including an AI answer. If advisors can discuss those answers openly, the client has less reason to conduct a separate conversation with a machine and keep the result to themselves.
Third, give the team a sensible way to handle volume. A fourteen-point email needs an acknowledgment and a decision about priorities. Some items can be dealt with in the meeting. Others may require research or a separate piece of planning. The client should know which is which. This is a service decision, and it deserves more thought than an irritated late-night reply.
Fourth, be willing to say when the AI has raised a fair point. Nothing damages confidence faster than an advisor brushing aside a question that later proves important. The strongest response may be, “I need to check that,” followed by a careful answer and, if necessary, a correction.
This is part of the larger question behind our AI Adoption Project. We set it up to examine how firms evaluate, implement and govern AI in practice. Future Proof gave us plenty to investigate on the firm’s side of that equation. The TRAP story is a reminder to include the client’s side.
I suspect we will hear many more stories like this one. Some emails will be absurd. Some will be irritating. A few will expose weak work. Many will come from good clients who are doing what they have always done: trying to understand a decision that matters to them, using whatever help is available.
The advisor in the podcast did something deceptively simple. He left the email where it belonged and waited to speak to the person who sent it.
That seems a good place to begin. Then the advisor has to be ready for the conversation: open to the useful question, alert to the faulty assumption and able to explain the work. Future Proof showed how much technology may change an advisor’s day. This is how it may change the next client meeting.
