AI can manage a family’s wealth. It cannot decide what the wealth is for
Delegating the work to artificial intelligence without setting goals could introduce inefficiencies
WITHIN a few years, artificial intelligence will take over much of the work in managing family wealth.
Consolidating accounts across entities, checking documents against the law, drafting and monitoring agreements, producing reports on every holding – tasks that occupy accountants, lawyers and family office staff are being automated.
Singapore and Hong Kong, where governments have actively courted family offices, will be among the first places to feel the change.
Consider a typical family that owns three operating companies and a portfolio managed by its single family office in Singapore.
A reporting system pulls figures from all of them every morning. Each company is ahead of budget, and the portfolio is performing better than its benchmark. By every measure on the screen, the year is a success.
Yet, the family has agreed that it will buy out two cousins and that the founder may start a new venture. Neither is communicated to the system; nothing says that the next three years must produce cash to fund these decisions.
Each company owned by the family reinvests its profits, as a well-run company should, and the system reports that this is going well because it was not made aware of new requirements.
The family office uses an investment platform that rebalances the portfolio towards its benchmark. The model finds regional property underweight and adds to it. The family’s largest company is a property developer.
Seen from the portfolio’s perspective, the allocation is balanced. Seen from that of the family, its exposure to one market has just gone up disproportionately.
Then a tax tool optimises the group’s structure. It works from the companies’ figures and assumes, as the companies do, that profits are reinvested. Corporate tax falls.
It does not consider tax on dividends because nothing the tool was given says the owners intend to withdraw cash. The plan is optimal for the companies; it is expensive for the family members when they are hit by their tax bill for taking out cash for the buyout and new venture.
Each answer provided by the system is correct. Each system optimises what it was given: the companies’ budgets, the portfolio’s benchmark, the companies’ tax. Each receives its instructions through a different channel.
None of them knows what the owners need, and the whole infrastructure drifts away from the family’s goals.
Compliance, relational, perspective
One way to see where this gap lies is through what I call “the three functions of ownership”: compliance, relational and perspective.
Compliance ensures all aspects of ownership, such as tax and regulatory rules, abide by the law. The law is codified, so much of this task will be automated by AI.
Relational is what the owners agree among themselves: the direction for their wealth, its limits, who acts on their behalf and what is reported back to them. Drafting and monitoring such agreements will also be largely automated, but the agreement itself remains a decision only the owners can take.
Perspective carries the agreement through. It gives each significant investment and business holding a mandate, assesses the strategy proposed to meet it, takes important decisions back to the family and tracks the results.
It does so with the family’s whole wealth in view, optimising within the limits that compliance sets, and, most importantly, reaches the organisation through one channel: the person leading the wealth on the owners’ behalf.
In the hypothetical family above, the chain breaks between the relational and perspective functions.
The owners have an agreement, but it was not decided and written down in the language of perspective. There is nothing to carry through, so each system optimises against the goals of a company or a portfolio instead.
Decisions still rest with the family
Today, an experienced finance director or family office head who knows what the owners want might catch all three concerns: question the reinvestment plans, notice the added property exposure, ask about dividend tax.
AI can do the same analysis, faster and across every holding, but only if the owners’ requirements have been established. Without that, automation does not close the gap. It makes the gap more consequential.
Closing the chain starts with a clearly communicated agreement. The family decides that they need cash over the next three years for the buy out and the venture, that their investment portfolio should factor in their property company, and that their taxes should be optimised for taking money out.
The person leading the wealth then turns that into instructions every system can work from.
Each company gets a mandate for what it must pay out, so its budget includes the cash the owners need.
The investment platform gets a limit on property allocation for the whole wealth, so owning a property developer does not unbalance the portfolio.
The tax tool is told that cash will be paid out, so dividend tax is part of the plan.
With that in place, AI can do what it does best. It flags the conflicts before the budgets are approved, before the platform rebalances and before the tax plan is adopted. The owners’ time is then spent on the decisions that are theirs.
For advisers, the shift is just as large. Much of what is billed today in compliance and relational work will be done by machines. The value that remains lies in helping owners define what the AI should serve.
AI will make it ever easier to produce information, assess alternatives and carry out decisions. None of that establishes what the wealth is supposed to bring the family. That remains with them, and carrying it through requires a function of its own.
The writer is founder of H S Nielsen, which advises owners of complex wealth structures
The commentary is based on the writer’s own experiences, observations and argument. AI tools were used for drafting in English and editing. The writer remains fully accountable for the commentary’s accuracy, originality and final form.
TRENDING NOW
What shattered Crystal Jade? How the culinary empire shrank from 120 to just 9 outlets
Thai brand Dr Pong takes T-beauty ambitions to Singapore as part of South-east Asia push
As Indonesia picks new vice-president to succeed Gibran, investors see risks and opportunities
Nvidia’s Jensen Huang fallout is a gift to billionaires Ambani and Adani