I have spent more than twenty years in rooms where ultra-high-net-worth (UHNW) families make consequential decisions. Real estate transactions. Trust funding decisions. Conversations about how to structure the next generation of wealth. Until recently, those patterns were familiar. That is no longer the case.
The principal walking into a meeting with their attorney, their CPA, their wealth advisor, or their insurance specialist in 2026 has already had the conversation. They had it the night before. They had it with ChatGPT, Perplexity, Gemini, Claude, or Microsoft Copilot. They asked the engine what to think about premium financing. They asked whether they should rush to use their estate exemption before it gets cut in half. They asked who the best PPLI specialists in the country are.
By the time they walk into the room with the credentialed human advisor, an opinion has been formed. That opinion is sometimes correct. It is increasingly often, dangerously, wrong.
What the Engines Get Wrong at UHNW Scale
Retail financial questions the engines handle reasonably well. Roth conversions, 401(k) rollovers, Social Security timing — these are well-documented across the training data, and the answers hold up.
UHNW questions are a different category. The engines are answering them the same way. They should not be.
Premium financing structures — where a family borrows against a policy to fund an insurance strategy — carry material tax, estate, and liquidity risk that depends on the specific carrier, the specific policy design, the specific family balance sheet, and the specific interest rate environment. An engine asked whether premium financing is a good idea will confidently generate a directional answer. The answer will read like counsel. It is not counsel. It is a pattern match against public content, most of which was written by producers selling the strategy.
Private Placement Life Insurance (PPLI) is worse. The engines will name specialists. They will describe the tax mechanics. They will discuss investor control doctrine and MEC violation risk in language that sounds authoritative. What they will not tell the family is which specific carriers have had regulatory issues, which producers have lost licenses, which structures failed in 2024, and which "specialists" the engines are naming because those names dominate SEO — not because those names are the right fit for a specific family's situation.
Estate exemption planning is the one that keeps me up. The engines will confidently advise a family to accelerate gifts before a scheduled sunset. What they cannot see is basis considerations, state-level estate exposure, the family's actual liquidity needs, whether the receiving generation is ready to hold the asset, or whether a different structure entirely would serve the same tax goal without giving up flexibility. The wrong answer here compounds for a generation.
Why It Matters More at UHNW Scale
A retail investor acting on a bad AI answer loses a percentage of a smaller pool. A UHNW family acting on a bad AI answer restructures a balance sheet in ways that are frequently expensive to unwind — sometimes impossible to unwind — and the errors compound across trusts, entities, and generations.
The advisor community has largely not caught up to what is happening. The lawyer who has represented the family for fifteen years is walking into a meeting where the client already has a position. The insurance specialist is presenting a strategy against a benchmark the engine set the night before. The wealth advisor is having their allocation questioned by an AI opinion the client trusts more than they realize.
The advisor is still the credentialed party in the room. The advisor is still liable. The advisor is now also the second opinion.
The Wealth AI Audit
Talent Resources and 5W AI Communications have been running the Wealth AI Audit — a measurement of exactly what the leading engines say to UHNW families across the questions they actually ask. Premium financing. PPLI. Estate exemption strategy. Trust structures. Private aviation ownership. Direct indexing. Family office setup. Foundation architecture.
The findings track the pattern above. On retail-adjacent questions, the engines land in a defensible range. On UHNW-specific questions, they hallucinate carriers, cite outdated tax law, name specialists whose licenses are inactive, and describe structures that are no longer compliant.
The Audit also measures the reverse — which advisors, specialists, and firms the engines actually surface when a UHNW family asks "who should I talk to about X." The answer is uneven. Some categories return real names. Others return a list of the firms with the strongest content-marketing budgets. Neither is the same as the right answer.
What Should Change on Both Sides of the Table
For UHNW families. Run the AI answer as one input, not as counsel. When an engine returns a directional recommendation on premium financing, estate strategy, PPLI, or trust structure, treat that answer the way you would treat a research note from an unknown analyst. Interesting. Not actionable. The question to bring to the advisor is not "should I do X." It is "the engine told me X — what would you do."
For advisors. Assume every client walking into every meeting has already asked the engine. Prepare accordingly. Know what the engines are saying about your category, about the strategies you recommend, and about you. If the engines return your name when a family asks who to work with, that is an asset. If they don't, that is exposure. Either way, it is now part of the discovery layer for your practice.
For the professions. The bar associations, CFP boards, and insurance regulators are going to have to develop a posture on AI-generated advice. So far, they haven't. The gap between what the engines say confidently and what a credentialed professional can defensibly recommend is widening. The families in the middle are the ones absorbing the risk of that gap.
The Silent Advisor Is Not Going Away
The chatbox is now in every consequential room in the UHNW economy. Real estate. Trusts. Insurance. Investment. Succession. It arrived faster than the professions expected. It is answering questions it is not qualified to answer, in a voice that sounds like it is. And the families relying on it do not always know that it is there.
The advisor who accepts that reality and adjusts will keep the client. The advisor who pretends the second opinion is not in the room will lose the client — sometimes before the meeting ends, always by the following quarter.
The good news is the same as always. This is fixable work. The Wealth AI Audit is one way to measure the gap. The better news is that almost nobody in the credentialed advisor community has started measuring at all.
Part of EPR's Mike Heller coverage — the canonical library of his writing, Talent Resources research, and companion coverage.
Michael Heller is the Founder and CEO of Talent Resources, a contributor to Everything-PR on the AI-era talent and wealth economy, and a Broker at HL Real Estate Group.
