Don't automate the workaround.Fix the Client File.

84% of the senior wealth leaders modelled in a new industry study want to rebuild their operating model for AI, and 7% are doing it. What usually sits between the two is a workaround: people assembling each client's picture by hand, from several systems, every time it is needed.

Put AI on top of that workaround and it runs faster while staying in place. The fix is one client record that the work runs through.

The figures come from Hidden in Pl(AI)n Sight, HCLTech study of AI in wealth management, released on 29 September. Its 1,066 respondents are AI personas modelled on senior wealth leaders across 17 markets, designed with input from the sponsor's own domain experts. The findings are directional, not a census.

The report is worth reading because it matches what we hear. We speak with private banks, family offices and independent advisers daily, and the pain points in this report are the ones they describe to us.

Where does a wealth firm's AI advantage come from?

From what the firm already owns. Asked to rank eight sources of sustainable advantage for a world in which AI is table stakes, respondents put proprietary client and behavioural data first, at an average rank of 2.1. Cloud infrastructure depth came last, at 7.0.

Distribution and trusted client relationships came second. Foundation model partnerships came seventh. Every firm can rent the same AI model and procure the same cloud. Decades of knowing how clients behave has to be earned.

The middle of the ranking matters too. Orchestration across vendors and ecosystem partners came third, ahead of owning a proprietary core platform in fifth. The advantage comes from making existing systems work together.

Where is value leaking in wealth management?

In fragmented client data. Asked to name the biggest and smallest sources of value lost between an insight and the action it should produce, respondents put fragmented client data and gaps in the 360-degree view first, at a net score of +15.3. Manual reconciliation of data across silos came second, at +11.5. Relationship managers switching between systems to see one client came third, at +11.3.

Read together, they are one problem seen from three chairs. The data team sees fragmentation. Operations sees reconciliation. The adviser sees five screens and a client waiting.

A single client view, in plain terms, is one current and consistent picture of a client's holdings, history, preferences, obligations and open actions, available without anyone assembling it by hand. Most firms do not have one. They have the parts.

The most visible failures ranked last. Errors in trade processing and settlement scored −18.0, and delayed or rework-heavy regulatory reporting scored −13.8. These incidents trigger escalations and board papers, yet respondents were confident the value goes elsewhere. The quietest failure is where value leaks.

Why doesn't adding AI fix fragmented client data?

Because AI combines fragmented data afresh every time someone asks, and keeps nothing it resolves. The stitching gets automated and stays where it was.

Modern AI can read the CRM, the portfolio system and the inbox, and combine them well. Conflicts between systems get smoothed over in each answer while the source records stay as they were. Two advisers can walk away with two different pictures of the same client.

An assistant rebuilds the client view each time. An operating system keeps one.

The report also shows where AI has gone so far. 34.5% of respondents have AI executing risk monitoring and alerts today. Only 8.0% have it executing proactive client communication. AI has gone where deployment is easiest, which is rarely where client value is created.

What does an AI operating model reset change?

It changes who does the work. Today, skilled people operate software: they gather data, reconcile it, prepare the meeting pack and chase the follow-up. In a reset operating model, AI carries out that standard work, and people take charge of what only people should own: judgement, exceptions, client relationships and accountability.

That makes the reset a redesign of work. Most core systems stay where they are: custody, portfolio management, planning. The client record is the exception. Every piece of standard work starts and ends there, so it has to sit where the work happens. Leave it outside, and the stitching moves somewhere new.

What does a wealth firm need in place before AI can run client work?

Five conditions, in this order:

1. One trusted, complete view of the client that AI can act on as well as read.
2. Defined permissions, so AI works within the firm's own governance: what it may access, what it may do, and what it must escalate.
3. Human oversight where it matters, with a named owner for every step that produces a regulated outcome.
4. A clear record of actions and decisions, designed in from the start.
5. The client record at the centre of the work, connected to custody, portfolio and planning systems.

Supervisors already expect much of this. In the UK, the FCA's Consumer Duty and the Senior Managers and Certification Regime mean accountability stays with the firm and its senior managers when a task moves to a system. In Switzerland, FINMA Guidance 08/2024 sets out supervisory expectations on governance, accountability and explainability for institutions using AI.

Oversight is cheapest when it is designed in.

Why is mass-affluent advice the AI prize most wealth firms are missing?

Because the report ranks it the most valuable AI revenue pool in wealth management, and few firms are prioritising it. Near-zero-cost advisory for the mass affluent tops that ranking. Mass affluent clients cannot carry the cost of a full adviser relationship today. That changes when client insight does the preparation work, extending coverage without adding headcount.

The urgency is missing. Only 11.8% of respondents say new revenue is their leadership's priority for AI. The inability to scale personalised advice to mass-affluent volumes scored just +3.8 as a source of lost value.

Coverage cannot be extended down-market on five screens and a spreadsheet.

What should an AI operating model change for clients?

The test of an operating model is what the client notices.

The client no longer has to repeat themselves. What they said in the last meeting is there in the next one. They hear from the firm before they have to ask, and proactive contact, the work only 8.0% of respondents have AI executing today, becomes routine. Their adviser spends the meeting on judgement. And if anyone asks why they received the advice they did, the firm can show them.

This is the problem WealthAi is built around. The WealthAi Client File brings together what a firm knows about each client from the systems it already runs. WealthAi Agents use that record to carry out standard work, such as meeting preparation and follow-up, within permissions the firm defines and with a clear record of actions and decisions. The client data, and the advantage it creates, stay with the firm.

None of that depends on a better AI model. It depends on one client record that the work runs through, with people in charge of what matters.