Home › Insights › Wealth Management › AI in Wealth Management: Your Client Is Already Advised
Wealth ManagementAI in Wealth Management: Your Client Is Already Advised
AI in wealth management is no longer an adviser-only story: clients use it first. See what European firms must do on governance, supervisors and the AI Act.
Picture the start of a client meeting. Before you have opened the file, the client slides a printout across the table. It is a critique of their portfolio, written overnight by a chatbot, and it is confident, fluent and not entirely wrong.
This is no longer unusual. Over a quarter of UK advised clients now use AI tools to support advice decisions. The AMF’s 2025 barometer found that 11% of French people consult AI before investing, ahead of social media (6%) and influencers (4%). Most firms have spent two years asking whether to adopt AI in wealth management. The more useful question is what to do now that the client has done so first. Our argument is that this is not a tooling project. It is a change in whom clients trust, and in what supervisors will hold you accountable for.
AI in wealth management: advisers have moved too
On the firm side, the debate is largely over. EY’s 2025 survey of 100 wealth and asset managers found 95% had scaled generative AI across multiple use cases, and 78% were already exploring agentic AI. In the UK, adviser AI use rose from 43% to 74% in twelve months (intelliflo, 2026). FINMA found roughly half of around 400 Swiss financial institutions use or are piloting it.
A word of caution on the numbers. The intelliflo figure rests on 209 advice professionals, so treat it as directional. The FCA’s survey of 4,100+ firms is more representative, and it shows an uneven picture: about 14% of small firms, 38% of medium firms and over half of large firms plan near-term AI use in advice processes. The mid-market is where the decision is still live, and where the economics are most attractive. Advisers report losing roughly three working days a week to admin that produces neither revenue nor relationships. Tax-return analysis dropping from an hour to three minutes is the sort of gain that adds capacity without adding headcount.
The client arrives pre-advised
Among UK clients who use AI for advice decisions, 44% wrongly believe the output is regulated, 38% think they can base an investment decision on it alone, and 32% mistakenly expect FSCS compensation. Forrester projects that more than half of under-50 consumers seeking advice will turn to generative AI rather than a licensed professional.
Advisers describe the result with some feeling. It is rather like having a frightfully well-read houseguest nobody invited: full of opinions on the portfolio, accountable for none of them, and showing no signs of leaving. Clients now arrive with AI-generated critiques that have to be discussed, not dismissed. To be fair to the evidence, this comes from adviser accounts and trade press, and no controlled study yet measures it.
The underlying asymmetry, though, is real. A chatbot has no duty of care, no suitability file and no compensation scheme behind it, yet clients increasingly treat it as a co-adviser. Firms that ask openly about AI use, correct its errors with evidence, and position the human as the accountable layer can turn this into trust. Firms that ignore it will find clients doing the challenging for them.
The wealth is moving to an AI-native generation
The demand engine is generational. Cerulli (June 2025) estimates $124 trillion will change hands through 2048, with Millennials receiving the largest share. Capgemini’s research finds 30% of next-generation HNWIs under 59 expect an inheritance by 2030, 63% by 2035 and 84% by 2040. These are model-based estimates, so read them as direction, not precision.
This cohort expects digital parity. Advanced digital capability is in next-gen HNWIs’ top four expectations of wealth firms, and 82% of Gen Z and Millennial generative AI users have already used it for financial advice (Credit Karma). But one proposition will not serve everyone. Inheritors tend to be sceptical but curious and want continuity and estate structuring. First-generation entrepreneurs are AI-optimistic, fact-check their advisers and want speed and direct deal access. Mass-affluent professionals face the highest substitution risk and suit a low-touch hybrid model.
Supervisors are enabling AI, on conditions
No major European supervisor opposes AI in advice. The FCA is actively enabling, with its AI Lab and live testing, and its July 2026 Mills Review concluded that no AI-specific rulebook is needed. It asks firms instead to place each use case on a five-role autonomy spectrum. The CNMV is more cautious. Its Working Paper no. 96 found that unstructured prompts produced calculation errors and hallucinations on IBEX 35 stock selection, while prompts anchored in official filings improved accuracy markedly. FINMA is technology-neutral, BaFin ties AI to DORA, and ESMA applies MiFID II’s best-interest duty in full.
The positions converge on one test: AI is acceptable where a named human firm stays accountable, acts in the client’s best interest, and can evidence its governance. Our readings of the Mills Review and the CNMV–FCA comparison go deeper. In short, the CNMV treats the human as a verifier, while the FCA treats the human’s position as a design decision.
The AI Act is in force, and parts of it already bite
The EU AI Act has been in force since August 2024, so an advice firm is not waiting for a start date. The bans on prohibited practices and the AI literacy duty have applied since February 2025, and the obligations on general-purpose AI models since August 2025. Staff training on AI is a live obligation today.
The high-risk regime is the part that moved. It covers credit scoring and life and health insurance pricing, and anything used to assess client eligibility deserves a careful look. The original start date was 2 August 2026. The Digital Omnibus, which entered into force on 27 July 2026, defers Annex III obligations to 2 December 2027 (and Annex I products to 2 August 2028). That is a planning horizon, not a reprieve. Classifying systems, documenting them and building human oversight takes months, and supervisors are not waiting: BaFin’s DORA-based guidance already applies, and the FCA, CNMV and FINMA all expect evidenced governance now. Firms that read the extension as permission to wait will arrive at December 2027 with the same backlog.
What wealth managers should do now
- Inventory and classify every AI system. Map each use against the high-risk list and record your reasoning. Do it now: the deadline has moved, but the work has not got shorter.
- Add AI to the fact-find. Ask clients what AI they use, script responses to AI-sourced claims, and document corrections in the suitability file. The FSCS misconception makes this a conduct and litigation issue.
- Set an autonomy ceiling per function. Cap anything that touches a client outcome or moves money below the approver level, and write down why before a supervisor asks.
- Sequence by return. Meeting transcription and report drafting first, agentic onboarding second, client-facing chatbots last and only with disclosure and human escalation.
- Ground the tools. Use retrieval anchored in official sources, keep client data out of general-purpose chatbots, and keep immutable logs of inputs, outputs and human decisions.
- Build next-generation engagement early. Digital reporting, scenario modelling and alternatives access for entrepreneurs, aimed at the 2030 to 2035 inheritance peak.
- Talk to your supervisor. The FCA AI Lab, dialogue with the AMF and CNMV, and FINMA’s technology-neutral stance all offer ways to de-risk flagship deployments before they go live.
In a nutshell
The dividing line in European wealth management is no longer between firms that use AI and firms that do not. It is between firms that can show a client, and a supervisor, exactly where the human decides, and firms that cannot. Clients already hold a free second opinion with no liability attached. What the adviser still has is accountability, which a chatbot cannot supply. Firms that make it visible, measurable and well governed will keep the relationship. In an AI-assisted market, the accountable human is the product.
Further reading
- On the supervisory view, FCA: our approach to AI.
- On firm adoption, EY: GenAI in wealth and asset management.
- On client behaviour, IFA Magazine: clients using AI to support advice.
- On generational transfer, Capgemini: World Wealth Report.
Related on our site: GenAI: from pilot to production in wealth management, Customer outcomes: winning wealth in a digital world, Beyond products and technology in wealth.
Further reading