AI Will Make the Mass Affluent Worth Advising

Financial advice has a strange distribution problem.

The people who could benefit from it are not always the people an adviser can afford to serve. A household may have meaningful savings, a pension, an inheritance, a mortgage, and five decisions it does not feel qualified to make. But if the revenue from that relationship does not cover discovery, documentation, compliance, portfolio work, and ongoing service, the traditional model struggles.

That is why so much advice remains concentrated at the top of the wealth pyramid.

The technology debate usually starts in the wrong place. It asks whether AI can replace the adviser. I think the more useful question is whether AI can make a good adviser economically available to many more people.

AI will not make human judgment free. It can make the work surrounding that judgment dramatically cheaper.

That distinction could turn the mass affluent from an underserved segment into one of the most important growth markets in wealth management.

The current model has a capacity ceiling

The UK market makes the constraint visible. The FCA's 2025 survey counted roughly 31,000 advisers serving 4.1 million retail clients and about £1 trillion of assets under advice. It described a typical adviser as serving around 150 clients, with approximately £250,000 of assets and £2,000 of revenue per client.

Those are market aggregates, not universal adviser economics. But they show the constraint. Scarce adviser time is consumed by collecting information, preparing meetings, drafting follow-ups, documenting suitability, and keeping records current.

The same FCA analysis says only around 9% of UK consumers take financial advice. It also identifies about seven million adults holding £10,000 or more in cash savings who may be missing the long-term benefits of investing. The gap is not proof that all seven million need the same product. It is evidence that the current service model does not reach a large group with real financial decisions to make.

Separate FCA research found an association of up to 10% higher wealth in the years after regulated advice, while warning that the effect becomes less certain over time and is difficult to separate from selection bias. Advice is not magic. Access still matters.

AI changes the cost stack, not the duty

The first wave of digital wealth management reduced the cost of portfolio construction. Model portfolios, automated rebalancing, and digital onboarding made investment management cheaper to deliver.

It did less to reduce the full cost of advice.

Advice is not just an allocation. It is a sequence of work: understand the household, organize its data, identify the decision, compare options, document the rationale, communicate clearly, monitor what changes, and intervene at the right moment.

AI can compress that sequence.

Before a meeting, it can assemble a permitted client brief from existing records, flag missing information, summarize recent activity, and prepare questions. During the workflow, it can retrieve policy and product information, draft analysis, and preserve source references. After the meeting, it can prepare notes, actions, and a client communication for review. Between meetings, it can monitor for defined events that deserve human attention.

Morgan Stanley offers a useful public example of this operating pattern. In 2024, the firm said 98% of its financial-adviser teams had adopted its internal AI assistant. It also launched a meeting tool that, with client consent, generates notes, identifies action items, drafts follow-up communications, and saves a record into Salesforce. That is a company-reported adoption figure, not an independent measure of client outcomes. The important point is the workflow: the AI does not replace the relationship. It reduces the administrative cost around it.

This is the same reason I believe the first banking AI winners will be invisible. The value will often appear as a prepared adviser, a faster answer, a timely intervention, or a cleaner record. The client does not need to watch the model work.

Mass affluent does not mean low complexity

One mistake would be to treat this segment as a simplified version of private banking.

The mass affluent often has fewer assets but no shortage of complexity. A client may have an employer pension, concentrated company stock, aging parents, school costs, insurance gaps, and tax consequences spread across different systems. The balance sheet may be smaller. The decisions are still connected.

That is why a chatbot alone will not close the advice gap. A chatbot can answer a question. Advice requires context, continuity, and accountability.

The right model is a service ladder.

At the first level, AI can provide education and navigation: explain concepts, surface relevant information, and help a person frame the decision.

At the second, a governed system can offer targeted or simplified support within a clearly defined scope.

At the third, an adviser takes responsibility for decisions that are material, ambiguous, emotionally difficult, or outside the system's authority.

The FCA is moving in this direction through its work on targeted support and simplified advice. Its 2026 wealth-management survey also found that 13% of respondent firms were using in-house or third-party AI tools, rising to 45% when firms considering adoption in the following twelve months were included. Those figures describe reported use and intention, not successful deployment. They show where the market is looking.

A model that can actually scale

If I were designing an AI-enabled mass-affluent proposition, I would build around five principles.

1. Standardize the preparation, not the person

Data collection, document classification, meeting preparation, and recordkeeping can follow consistent processes. The recommendation should still reflect the client's goals, constraints, and circumstances.

Efficiency should remove repetitive work, not flatten people into a segment label.

2. Move from calendar service to event service

Traditional advice is often organized around an annual review because adviser capacity is limited. AI makes continuous monitoring possible.

The system can identify a defined event: excess cash, a maturing product, an unusual portfolio concentration, a contribution gap, or a change in circumstances recorded by the client. The adviser engages when there is something worth discussing.

That is better service and better economics.

3. Reserve people for consequential moments

Human time should go where judgment, empathy, negotiation, or accountability matters most. The technology should prepare the evidence and reduce administrative friction.

But the escalation design has to be real. As I have argued before, "human in the loop" is not a governance model. The person needs context, authority, time, and a specific reason to intervene.

4. Measure outcomes per household

The business case cannot stop at hours saved. Measure the cost to serve, clients per adviser, response time, plan completion, follow-through, complaints, corrections, retention, and evidence of improved financial behavior where it can be measured responsibly.

More conversations are not automatically better advice.

5. Keep a complete decision record

Every recommendation should preserve the relevant inputs, source material, applicable policy, system contribution, human review, and final action. That protects the client and gives the institution a way to improve the service.

Personalization without traceability is not scalable advice. It is scalable risk.

Trust is the limiting factor

There is real consumer interest, but it is not unconditional. The FINRA Investor Education Foundation reported in 2025 that 20% of US adults surveyed were interested in receiving financial advice from AI. The FCA's 2026 wealth-management work similarly reported that one in five UK adults were open to AI making financial decisions for them.

Those findings do not mean consumers want an autonomous portfolio manager with unlimited authority. They indicate an addressable group willing to consider new delivery models.

Trust will depend on boundaries. Clients should know when they are interacting with AI, what data it uses, what it can do, when a person reviews the result, and how to challenge a decision. The institution remains accountable. A lower service cost cannot mean a lower standard of care.

The direction of travel is already visible in the bank operating model of 2030: more clients served through systems that prepare, monitor, and execute defined work, with people concentrated on exceptions and relationships.

We should not describe that future as advice without advisers.

It is advice with much less administrative waste.

For decades, wealth management rationed human attention because the machinery around an adviser was expensive. AI can change that machinery. If institutions redesign the service rather than add another tool, millions of households can become economically viable to serve well.

The mass affluent was never short of decisions. The industry was short of a delivery model that could afford to help with them.

AI can build that model. The adviser still has to earn the trust.