The profession is asking how AI can make advisers more productive. What happens when it makes customers more capable?
This week, financial planners gather in Windsor to “turn the page on established thinking”.
Here is the page I would turn.
Imagine your next prospective client arriving on a video call. They have organised their financial information, explored retirement scenarios, compared charges and prepared questions about your recommendations. They have also asked an AI to challenge their assumptions—and yours.
They are not starting from a blank page. They are starting with homework.
What are you going to do in that meeting that has not already been done? What will you check, discover, resolve or take responsibility for that makes a meaningful difference?
And how will you charge for it?
That is the conversation I want our profession to have.
I have felt increasingly outside the conference conversation in 2026. My concern is that a message challenging the established business model is too easily heard as anti-adviser, or dismissed as maverick. I cannot speak for organisers’ reasons for choosing their speakers. But I can explain why this question deserves a platform.
Customers have AI too.
The CISI’s published conference announcement includes AI tools for planners, behavioural science, storytelling and lifestyle financial planning. These are worthwhile subjects. The PFS’s November programme also addresses AI and changing client needs. It would be unfair to pretend these institutions are ignoring change altogether.[1][2]
My challenge is more specific: will the discussion go far enough to examine what happens when customers need less of the service around which the profession has organised its income?
There are two sides to the AI story. On the firm’s side, it can reduce the cost of producing a service. On the customer’s side, it can reduce the amount of that service someone needs to buy.
Celebrating the first while overlooking the second is a dangerous business strategy.
The provocative question is: what happens when AI can do everything you do for free?
We do not have to accept that as a literal description of today’s technology to take the threat seriously. AI does not need to replace every professional task. It only needs to make enough previously chargeable work available at little or no direct cost for customers to reassess the price of the package.
A business can remain useful while its traditional pricing stops making sense.
“But AI makes mistakes.”
Yes. So do people.
The relevant comparison is between the real alternatives available to a particular person, including their costs, biases, safeguards and consequences when something goes wrong.
Recent research makes easy slogans difficult to defend. A 2026 study by Ylva Baeckström and Roman Matkovskyy found that professional advisers projected their own portfolio preferences onto clients. AI recommendations also showed biases that varied with the model and prompt. The AI portfolios were more conservative and produced lower modelled long-term wealth, but human advisory fees eroded much of the human portfolios’ advantage.[3]
That was a portfolio experiment, not a verdict on the entirety of financial planning. It does not establish that humans always make more mistakes, or that AI is always superior. It does show why “human” cannot simply be treated as a synonym for “unbiased”, and why fees belong inside the comparison.
Research described by MIT Sloan also found that AI could encourage useful saving and investing behaviour, while struggling with changes such as unemployment. More structured prompts improved its performance. These were simulated outcomes, not proof of lifelong results for real clients.[4]
The evidence calls for evaluation. “AI makes mistakes” cannot do all the work of defending a business model.
We also need to be honest about who receives human advice.
The FCA’s Financial Lives 2024 findings showed that only 9% of adults had taken full regulated advice in the preceding 12 months.[5] That does not mean the other 91% had no access to a planner. Some did not need advice that year; others chose different support.
Nevertheless, for someone unable or unwilling to pay for a planner, the alternative to AI may be an unanswered question, a sales pitch, a social-media tip or another year of avoidance. Free guidance and other support also exist. The comparison must reflect what that person would actually use.
We cannot assess every accessible tool against an ideal professional relationship that the person does not have.
Then comes the reassuring response: “AI will do the administration. We will focus on judgement and behaviour.”
I have news for you. Customers can ask AI for help with those things too.
They can ask it to challenge a conclusion, examine trade-offs, rehearse a difficult conversation, identify inconsistencies or help them pause before acting. The quality will vary. An agreeable answer can reinforce a mistake. But the words “judgement” and “behaviour” do not, by themselves, establish a uniquely human service.
If those are your sources of value, demonstrate them. Explain what your involvement changes, how you recognise when the AI is wrong, and what responsibility you accept for the work you undertake.
A customer may value continuity, discretion, personal knowledge, practical coordination or a professional willing to stand behind a recommendation. Another may deliberately choose to delegate because they have neither the time nor the desire to manage everything themselves.
Agency includes the freedom to delegate. It also includes the freedom to take back control.
That brings us to ongoing fees.
The FCA reported in its 2025 review that about 80% of revenue from adviser charges related to ongoing services.[6] That makes the question commercially significant: how should recurring charges respond when a customer’s need for specialist advice is episodic?
Some people need substantial continuing work. Monitoring, coordination and availability can have real value. A quiet year does not prove that no useful work occurred.
But a continuing relationship is not, on its own, an explanation of a continuing fee. The service needs a clear purpose, an intelligible price and a credible account of what is delivered between major decisions.
For a customer whose needs are occasional, a project fee, a decision review or a modest support subscription may fit better. For someone with continuing complexity, a more substantial service may be appropriate. The model should follow the need.
Percentage charging deserves the same scrutiny. If a portfolio doubles, what work, responsibility or benefit increases alongside the fee? The answer may sometimes be persuasive. It should still be given.
There is a wider incentive problem here. Where revenue depends on assets under advice, there is a commercial reason to pay particular attention to assets that can enter or remain within that arrangement.
That does not make every adviser biased. It does make the structure worth examining.
A person’s most valuable next step may concern earning capacity, health, housing, caring responsibilities, relationships or time. An investment portfolio is only one part of a life. AI can help people explore a wider set of questions, although its own assumptions and commercial incentives also deserve scrutiny.
This is why I see a potential shift from advice to agency: people becoming more able to understand, question, choose and act for themselves, with professional help where it adds value.
It is a direction of travel to build for, not a claim that every customer has already arrived.
There are practical ways to respond now:
- Start meetings with the customer’s existing work. Ask what they have explored, which sources they used and what remains uncertain.
- Sell clearly defined contributions: testing assumptions, resolving complexity, coordinating action or providing accountable specialist advice.
- Match fees to the pattern of need, with clear choices between occasional and continuing support.
- Help customers keep a usable record of their own plans, decisions and next steps.
- Measure whether people leave more capable of acting, as well as whether they remain clients.
This is the direction I am building towards through the Academy of Life Planning: a person operating their own Total Wealth Plans system, with a Total Wealth Planner providing human support when needed. The design must also help people recognise when independent action is unsafe or specialist expertise is necessary.
Continuous capability. Episodic expertise.
The consequences reach beyond advisers. Paraplanning and compliance services may face pressure to demonstrate value as routine production becomes cheaper. Professional bodies will need to consider what competence means when clients can interrogate professional work themselves. Regulators face questions about accountability, consumer protection and the quality of the tools people actually use.
These functions will not simply disappear. Their work may change substantially. Some existing revenue models may not survive unchanged.
Why, then, does the debate so often feel more comfortable discussing efficiency than dependency?
My interpretation is that institutions find it easier to discuss innovations that strengthen their members’ current businesses than changes that might shrink parts of them. A session about producing reports faster offers an immediate benefit. A session asking why customers should keep buying those reports creates a harder conversation.
That is an incentive worth recognising, not proof of deliberate exclusion. I would welcome CISI and PFS putting the question explicitly at the centre of their programmes.
I am not anti-adviser. I want planners to have a future worth choosing, and customers to have choices worth making.
There is an express train approaching, sounding its horn. Discussing how to work more efficiently on the track is an inadequate response.
You will not hear me from the Windsor conference stage this week. But I hope you will take this question into the room:
When your customer has AI too, what will they still choose to pay you for—and how will you earn that choice?
Sources
- CISI: Financial Planning Conference 2026 agenda announcement.
- PFS: Festival programme announcement and speaker programme.
- Baeckström and Matkovskyy (2026), Financial advice behaviour: humans versus AI, Journal of Corporate Finance. Findings refer to a vignette-based experiment and modelled portfolio outcomes.
- MIT Sloan: AI financial advice is surprisingly good—especially if you ask the right questions.
- FCA: Financial Lives 2024 advice-use finding.
- FCA: Ongoing financial advice services, 2025 review.
