We use these tools. We respect them. An AI assistant is a real piece of equipment, and like any good piece of equipment, it does specific jobs well and other jobs badly. The difference between a client who is helped by AI and one who is hurt by it comes down to knowing which job is in front of them.
Here is the principle that sorts it. AI is excellent at helping you understand and prepare. It is unfit to make any decision that is hard to reverse, that crosses more than one part of your financial life, that turns on current law applied to your facts, or that depends on numbers it has never seen. Almost everything below follows from that one line.
Think of it as a traffic signal.
Green light: use it freely
There is a whole category of work where AI is genuinely useful and carries almost no risk, because none of it is a decision. It is preparation.
Use it to understand a concept before a meeting. If your estate attorney mentioned an irrevocable trust and you want to walk in already knowing what one is and how it differs from a revocable trust, AI will explain it clearly and patiently, as many times as you need, without making you feel slow for asking. The same goes for understanding how an installment sale spreads gain, or what a 1035 exchange is, or why a Roth conversion might matter. A companion piece explains why the tool is so good at sounding clear, and for learning a concept, clear is exactly what you want.
Use it to build the list of questions you should be asking. Describe your situation in general terms and ask what you should raise with your CPA, your attorney, or your wealth manager. The questions it produces will make your next meeting sharper and your professionals’ time more valuable. You walk in prepared instead of nodding along.
Use it to make sense of a document you already have. Paste in a section of a statement or a plan summary you received and ask it to explain the terms in plain language. You are not asking it to decide anything. You are asking it to translate, and it translates well.
Every one of these makes you a more informed, more fluent client. That is real value, and you should take it.
Yellow light: use it, then verify
The next category is useful with a guardrail. Here AI gives you a starting point, and the rule is that a starting point is all it is.
Rough scenarios belong here. Asking how a Roth conversion might affect your taxes in general terms, or what the trade-offs between two strategies tend to be, gives you a useful first sketch of the landscape. So does a back-of-the-envelope framing of a decision before you bring it to a professional. These first passes orient you. They help you ask better questions and follow the answers.
The guardrail is simple and it does not bend. Anything from this category gets checked by a person who can see your actual documents before it touches a dollar. The AI’s general sketch was built without your tax return, your trust, your account titling, or the current law as it applies to you. It is a reasonable picture of the typical case. Whether your case is the typical case is precisely the thing it cannot know. Treat the output as a question to bring to someone, and the yellow light stays useful.
Red light: do not let it decide
The last category is where the real money lives, and where AI has no business making the call.
Do not let it drive a decision you cannot reverse. Exercising stock options, surrendering a policy, completing an exchange, signing a structured sale: these are permanent. We told the story of an executive whose AI confidently approved an option exercise that triggered a $90,000 tax bill she never saw coming. The advice was accurate about the part the AI could see and silent about the part that mattered, and by the time the gap surfaced, the decision could not be undone.
Do not let it decide anything that crosses more than one part of your financial life. A tax move that has estate consequences, an insurance change that affects your investment allocation, a gifting decision that touches all four at once: these require someone who can see every corner at the same time. AI answers the corner you describe and has no view of the other three.
Do not let it make any decision that turns on your specific numbers or on current law applied to your facts. It does not have your numbers, and it cannot reliably confirm what the law is for your situation today. It will produce a confident answer anyway, which is the trap.
When a decision is irreversible, crosses pillars, depends on the law, or depends on numbers it has never seen, the AI’s role ends. The decision belongs to a person who can open your file and who answers for the result.
Why the line sits where it sits
Notice what divides the green light from the red. AI is built for the parts of your financial life that are general and reversible: learning, preparing, drafting, translating. Your money is made or lost in the parts that are specific and permanent: the actual exercise, the actual surrender, the actual sale, executed against your actual numbers, with consequences that cannot be taken back.
That dividing line is the exact line this firm sits on. We tell every prospect plainly what we will and will not do. Drawing the same kind of line around AI is the same discipline turned on a new tool: clear about where it helps, clearer about where it has to stop.
Used the way this article describes, AI will make you a more informed client, and a more informed client is a better client to work with. What it will not be is your advisor, because it has never seen your whole picture and it answers for nothing. That part of the work still belongs to a person, sitting across from you, accountable for how it turns out.
