The Chair Across the Desk

Two versions of this story have been coming at you, and you have probably met both in the same week.

The first says the profession is finished. Advice is about to be free. Somebody types their situation into a chat window at ten at night, gets back three tidy pages with headings and a suggested withdrawal rate, and wonders over Thanksgiving what exactly the one percent was for. The screenshot goes around. Underneath it, a stranger who has never once sat with a couple who disagree about when to retire explains that planning is a solved problem now.

The second version arrives by email four times a week with a demo link attached. The AI-powered practice. Prospects nurtured while you sleep, notes that write themselves, plans in seconds, a line on a chart that only points one way. Thirty days free, no card needed.

Both stories are built the same way. Each is the size of a decade, each is set somewhere out ahead of you, and neither one makes a single job on your desk any smaller. You could pick a side over dinner and Wednesday would look identical. The summary letters from last week’s two reviews are still unwritten. The quarterly letter is nine days past the date you set for yourself. The CPA you had coffee with in March still doesn’t know which of her clients to send you, because the note that would have told her never got typed.

Neither story helps with any of that. This book is about the letters.

Deal with the first story now, because it comes apart quickly, and it comes apart on specifics.

It cannot sit in the chair across from you. Not on an ordinary Tuesday, and certainly not on the hard days — the appointment six weeks after a funeral, where somebody arrives with a folder they have never opened and a set of decisions their spouse always handled. Nothing in that hour is a writing problem. You are watching how much a person can absorb today, working out what can wait until the fall, and saying a number out loud kindly.

It cannot hear the pause. Eleven minutes into a review, when the conversation is still about the kitchen renovation, and one of them says well, we’ve been helping our son a bit, and the other one looks at the table. That is the meeting. No form caught it and no questionnaire asked.

It cannot judge risk tolerance either, and that one is worth saying plainly, because scoring a questionnaire is exactly the sort of task software looks good at. A couple ticks moderate in a calm February. Moderate is what they are at rest. What you need to know is who they become at ten in the morning on a red day with the television on in the kitchen, and you learn that by sitting through one with them, or through enough of them with other people to know the tells.

It does not carry your registration. Your name is on the paperwork, the ADV or the U4 or both, and the examination letter comes to you. The duty you owe — fiduciary, best interest, whichever regime you sit under — is a legal obligation attached to a person. No software holds a registration, and none is going to be issued one. The suitability judgment is yours and stays yours, and the workflows in here are shaped so that it cannot quietly become anything else.

And it is not a source. That matters more in your work than in most, because it will sound exactly like one. Ask for this year’s contribution limit, a rule citation, a long-run average, and back comes a clean confident sentence in the same voice it used for the true parts. Chapter 10 takes this apart in detail. For now, put anything it hands you with a figure or a citation in it where you would put a headline a client forwards at midnight: possibly true, checked by nobody.

So: no. Not the advice, not the judgment, not the relationship, not the license. That is every reassurance this book has, and none of it needed a chapter.

The worry underneath it is not foolish, though. Something real is moving; it just is not a machine doing financial planning. It is that your clients now have somewhere fluent to ask at eleven at night, and what comes back will be organized, plausible, generic, and completely uninformed about the son they have been helping. It will also arrive in four seconds, which is roughly four days faster than most practices reply to anything.

That is the part worth watching, and it is a writing problem — which is convenient, because writing is the thing here that gets faster.

Now the true half, and it is not the reassuring one.

Something is changing. It is not the advice. It is the writing that surrounds the advice.

No exam you sat prepares you for the volume of it. You thought you were building a practice that plans and manages money, and you were, and you also built a thing that quietly produces a few thousand words a week. The agenda before the review. The summary letter after it, the one clients keep, the one that stops I thought we decided to wait. The CRM note that becomes the record of what you advised and why. The volatility letter, on the morning six meetings are already booked. The quarterly letter. The fee explanation for the hundredth time, slightly different every time. The cover note for the custodian paperwork, the RMD reminder, the short message that says nothing needs doing but here is what that headline meant. The introduction letter to the estate attorney, which does not get written, so that channel stays shut another year.

None of it is advice. All of it is what your fee buys, in the only place a client can see it. Almost all of it lands after seven in the evening, or it never lands at all.

That pile is words, and words are what these tools do well. Not the recommendation. The four paragraphs that explain the recommendation to somebody who half-remembers the meeting and will read it at the kitchen table with the television on.

Their weak points do not come up in the demo. The first is invention: a made-up figure arrives with the same calm delivery as a real one, and nothing in the writing marks the difference. The second is instinct. Left alone, these tools reach for comfort that is a compliance problem wearing a kind face — protected, guaranteed, markets always recover. And they know nothing about your fee schedule, your custodian’s real timelines, your firm’s written procedures, or the couple you met this morning.

So the workflows here hold on to the facts, the numbers and the decisions, and hand over nothing but the sentences. Wherever that split gets slippery, a WATCH OUT box spells out the check: numbers, promises, rules, about sixty seconds. Sixty seconds a draft is what the tool costs to run, and set against fifty minutes it is not a hard trade.

One sentence on the part that makes any of this workable in a regulated practice, and then the rest of the book will keep repeating it. The AI changed who types the first draft. It changed nothing about who reviews, who approves, or what gets archived.

Anyone with ninety households knows the shape of review season.

Four annual reviews in nine days, which is fine, because reviews are the part you are good at. Tuesday’s two go well. The couple at ten leave with a Roth conversion decided and long-term care parked until spring. The one at two brings up a job change that rearranges the next eighteen months, and you handle it in the room, well.

Then Tuesday evening you sit down to write the first summary letter. It takes fifty minutes, because you are not summarizing anything, you are explaining: what was decided, why the other option was set aside, what happens next and who does it. By the time you read it back it is the best thing you will write this month, and it will sit in a drawer in that house for years.

The second one you start at twenty to ten and abandon at ten past, half written, promising yourself first thing.

First thing is Wednesday, and Wednesday has two meetings of its own.

By Friday the letters have turned into a task that carries guilt, and guilt makes a task heavier, and heavier makes it later. Two go out the following week, thinner than they should be. One never goes at all, and the CRM note for that meeting gets typed three weeks later from memory: five lines, no rationale, nothing about the conversation that made the decision make sense. That is the file now.

In June that client calls and says he thought you had agreed to wait on the conversion. You know what was decided. But the record is five lines written from memory in the fourth week after, there is no letter, and this is suddenly a conversation about who remembers what, with somebody who has never doubted you before.

The failure there is not skill. You wrote a good letter on Tuesday night and it was clear and warm and correct. The trouble is that a good letter costs fifty minutes, and the calendar handed you four of them and two evenings.

The scope of this book, plainly, before you spend an evening on it.

It is a manual for turning one of the everyday chat tools — Claude, ChatGPT, Gemini, Copilot, whichever your firm has approved — loose on the written work of a small advisory practice. Review agendas and summary letters. Volatility letters, quarterly letters, the short note that goes out when news lands on somebody’s situation. Fee explanations, onboarding sequences, custodian cover notes, dictated debriefs turned into CRM-ready records. COI introductions and the one-pager a CPA can hand to somebody. Educational articles, a month of LinkedIn posts, seminar outlines, CE study. There are 150 prompts, numbered and grouped, and a library at the end that finds you the right one while a client is still on hold.

None of that library is aspirational. Each prompt was tested in Claude, ChatGPT and Gemini alike, and any that only behaved on one of them was dropped. What is left is wording that already worked, not a list of things AI is said to do for advisors.

What it is not: a planning tool, a research source, a portfolio engine, a compliance shortcut, or a stand-in for your firm’s written procedures and your compliance officer’s judgment. The recommendation is never its work, and no prompt in here invites it to try. Anywhere the tool comes within reach of something expensive — client data, performance language, testimonials, the record itself — a RULES CORNER box marks the line. Chapter 10 gathers every one of those lines into a single chapter, and that is the chapter to read before anything AI-drafted goes near a client. One rule outranks the rest of the book: your firm’s AI policy wins. If compliance has not approved the tool, the workflow waits until you are on one they have.

None of this needs you to be technical. It needs an email address, half an hour, and a browser tab you leave open. Chapter 2 does the setup. Chapter 3 teaches the one skill the rest depends on, and you have had it since the first time you handed work to a paraplanner: say who the client is in broad strokes, what is true, what you have already decided, what you want back, and what it should look like.

One note on scale before Chapter 2.

Your life stays as it is. What moves is the block between six and nine, and, a month in, most evenings in the week. The Promise Page opposite puts about eleven and a half hours on that, which is a real figure and not a free one. The first week will feel slower, the way anything does before your hands know it. By the third week it stops being a task you are doing and turns into how you write.

If two things stick, the book has paid for itself, and two is what most advisors take: the two-minute debrief that becomes the CRM note, and the summary letter that goes out the same afternoon. The marketing chapter never gets opened. Four hours come back. Nobody is grading you on the other ten chapters.

Nobody is coming for the chair across from you. Somebody has to sit in it, hear the pause at minute eleven, and carry the judgment afterward, and that will be a person with a license and a memory of what happened last time. What is actually on the table is far smaller than either version you have been sold, and a good deal more useful: the summary letter out the same afternoon as the meeting, ten minutes instead of fifty, and a Tuesday evening that is yours.

Half an hour of setup. Start there.