And why some clients will need a human in the room more than ever.
I want to say something that I think a lot of people in my industry are afraid to say out loud.
Artificial intelligence is going to replace a great many financial advisors over the next decade. Probably most of them. The advisors who run a tidy practice doing solid, competent work – gathering documents, building plans in commercial software, rebalancing portfolios on a quarterly schedule, screening for tax-loss harvesting opportunities, drafting client letters that sound like every other client letter – those advisors are in real trouble. Not because they are bad at their jobs. Most of them are quite good. They are in trouble because the things that make them good are the things AI can now do faster, cheaper, and at three in the morning.
I am not predicting this from a distance. I use AI every day in my own practice. Anyone in my industry who tells you the technology is overrated has not actually used it. It is not overrated. It is one of the most consequential tools to come into financial services in my career, and it is improving on a timescale that should make every advisor pay close attention.
So let me be clear about what I am claiming and what I am not. I am not claiming that the human advisor disappears. I am claiming that the bar for being worth what you cost is rising quickly and that the people who used to settle for a good advisor will, before long, get a better version of that good advisor for free, or close to it, from a chatbot. The gap that survives is between the great advisor and the AI. And great is harder to define than most of my colleagues like to admit.
What AI is genuinely good at – and what “good” advisors mostly do
If you sit down and list the actual tasks a competent advisor performs in a typical week, a sobering thing happens. Most of them are tasks AI either already does well or is about to. Pulling in a client’s documents and summarizing what is in them. Comparing two retirement scenarios. Calculating a Roth conversion. Drafting a quarterly letter. Researching a tax law change. Evaluating an annuity contract. Building a cash flow projection. Spotting a missed beneficiary designation.
None of this is glamorous work, but it is most of the work. And it is the work that, until very recently, justified a full time person doing it. That justification is eroding. Software that costs a client $30 a month can now do a credible first pass on most of these tasks, and within a few years it will do a near-final pass. The advisor who built a career on being the diligent middle layer between the client and the financial machinery is being replaced from underneath by a tool that does the diligent middle layer for free.
That is the bad news. Now the good news, which is also the more interesting news.
The clients AI cannot serve well
There is a category of client for whom AI will never fully be enough, and the reason has nothing to do with technology. It has to do with the structure of the problems they are trying to solve. In my experience, three characteristics tend to define this group. Many of the people I work with fit these characteristics. The clients who need a great advisor likely have all three.
First, complicated financial situations. I am not talking about a household with a 401(k), a Roth IRA, and a mortgage. AI handles that beautifully. I am talking about the business owner whose personal balance sheet is wound around an operating company, a holding LLC, a real estate entity, and a buy-sell agreement that has not been updated since the partner left. I am talking about the family that has wealth flowing across two generations, with trusts that were drafted in different decades by different attorneys with different assumptions. I am talking about the executive whose compensation includes restricted stock, performance shares, deferred comp, and a non-qualified plan that interacts with their cash flow in ways that change every year. AI can produce a remarkably good overview of any one of these pieces. Where it struggles is in the connective tissue, the place where the entity structure, the estate plan, the tax exposure, the family dynamics (always the gasoline on the fire!), and the operating reality of a closely held business all touch each other. That is where decisions actually live, and that is where the analysis is messy enough, and the data is incomplete enough, that you need a human who has seen this kind of mess before.
Second, a high regularity of consequential decisions. Some clients live a financial life with very few decision points. They save into their plan, they hold a diversified portfolio, they rebalance on a schedule, they retire on a date, and most of the work is just steady execution. They benefit enormously from a sound plan and a low-cost portfolio, and frankly, AI can carry a lot of that load. But other clients face a steady drumbeat of real decisions. Should I take the buyout offer or hold out for a better one? Do I exercise the options now or wait? Should we sell the second home or keep it? Do we lend to our son’s startup or write a check we cannot ask back for? Do I move the trust to a different state? Do I take the partnership stake? Do I retire in eighteen months or push it three more years? When decisions of this kind arrive every few weeks rather than every few years, you do not need a tool. You need a thinking partner, someone who knows your situation cold, who you trust, who you can call on a Tuesday afternoon and say, here is what I am turning over in my head, what am I missing? That kind of relationship is not something a chat window provides, no matter how clever the chat window gets.
Third, high consequences and costs attached to those decisions. A wrong move on a $40,000 401(k) contribution is forgivable. A wrong move on a $40 million liquidity event is not. A misread on the timing of a Roth conversion can cost a few thousand dollars; a misread on the structure of a business sale can cost seven figures and a strained relationship with a sibling. When the dollar amounts get large enough, or when the decisions become irreversible enough, the value of being right goes up, and the value of being wrong goes up faster. Clients in this position are not paying for information. They are paying for judgment under pressure, and they are paying for someone to share the weight of the decision with them. That is fundamentally a human service. It always has been. AI does not change it. If anything, AI raises the stakes, because the people on the other side of these transactions…the buyer, the IRS, the opposing trustee, the estate attorney…are using AI too, and the playing field at the high end is getting more sophisticated, not less.
What “great” actually means now
I have been thinking a lot about what separates the advisors who will thrive in the next ten years from the ones who will not, and a lot about the clients who will need them. It is not credentials. It is not technical knowledge. AI is a great equalizer on technical knowledge, and the playing field there is collapsing fast. There is no longer a scarcity premium added to knowledge. The advisors who will thrive are the ones who do the work AI cannot do, and that work has a specific shape.
It is the work of being a thinking partner before a decision, not just a report generator after one. It is the work of pushing back when a client wants to do something you believe will hurt them, and doing it in a way they can hear. It is the work of holding a conversation across years, remembering what the client said three Christmases ago about their daughter, and connecting it to what is being decided this Tuesday. It is the work of judgment in places where the data is incomplete, the stakes are real, and the answer is not in any model. It is, in a word, presence. The kind of presence that does not scale, cannot be automated, and is exactly the thing that the right kind of client will pay for as long as I am alive to provide it.
If you are a client of mine reading this, I want you to know I take this seriously. I am using every AI tool I can get my hands on, not to replace what I do for you, but to free up more of my time and attention for the part of the job that actually matters when your number is called. The diligent middle work…the research, the modeling, the document review, the first drafts…should be done faster and cheaper every year. That is good for you. The real work, the conversation that happens when something hard is on the table and you need someone in the room with you, is exactly where I want to be spending more of my time, not less.
And if you are reading this and wondering whether you have the kind of financial life that justifies a great advisor – whether the complexity, the decision velocity, and the stakes really warrant the relationship – that is a fair question to ask. For some people, the honest answer is no. A good chatbot, a target-date fund, and a disciplined savings habit will get them where they are going. For others, the answer is firmly yes, and the cost of being wrong about it is too high to leave to a tool that, however brilliant, has no skin in your game.
Knowing the difference is itself a financial decision. And it might be the most important one you make this year.
