The Best Time to Bring AI Into Your Firm, Business or Job Is When It Is Already Changing

Yesterday someone tried to sell me a lead list. I said no to the list, but one thing in the pitch stuck with me.

The three sample "opportunities" he sent were all accounting firms in the middle of a change. One had just joined a larger network. One had been acquired and was folding its systems into the buyer's platform. One was the second acquisition in a month for a firm stacking up offices. His point was that a firm in the middle of a change is rebuilding how the work gets done anyway, and that is exactly when it is willing to try something new.

He was right about that. And it is not just true of accounting firms. It is true of any business, and of any job.

Nobody changes how they work on an ordinary Tuesday

Every firm, every business and every job has a way of doing things, and most of it was never designed. It accumulated. The month-end checklist is a document someone made in 2019 and nobody has touched since. The follow-up on unpaid invoices lives in one person's inbox. The welcome email for new customers has a paragraph nobody remembers adding. The weekly report you send your boss takes two hours because that is how long it took the first time.

On an ordinary week there is no reason to touch any of it. It works well enough, and changing it costs time nobody has.

Then something changes. A merger. A new partner. A move to new software. A key person leaves. You start a new role, or get a new manager, or take on someone else's work. Suddenly the old way of doing things has to be written down, explained, moved or rebuilt, and the cost of changing it drops to almost nothing, because you are changing it anyway.

That is the moment to bring AI in. Not as one more project on top of the change, but as part of how the new way gets built.

The change you can schedule: the off-season

Most of us are not in the middle of a merger. But every firm in tax has a change it can count on, and it is coming in thirteen days.

October 15 is the last extension deadline. After that, the season is over, the pressure drops, and for a few months there is room to think. The off-season is the closest thing a tax practice has to a planned rebuild, and most firms spend it recovering instead.

If you are not in tax, you have one too. Every business has a slower stretch, and every job has a quieter month. That is your off-season, whether you call it that or not.

Here is a different way to use it. Spend the next few months getting ready for a better busy season. For tax, that means a better April 15, 2027:

• Write down the three jobs that hurt most this year. Not in general. The specific ones. For most firms it is the document chase, the organizer that nobody fills in, and the review backlog in the last two weeks. For a small business it might be the quote follow-up, the invoice chase and the weekly numbers. For a job, it is the part of your week you dread.
• Pick one and hand it to an agent. One job, built properly, with quality checks, tested on old examples with the identifying details taken out. If the agent can do last year's version of the job on old data, it can do next year's on real data.
• Build the rhythm before you need it. An agent that only runs when someone remembers is not a system. Set it to run on a schedule, so that when the busy stretch comes, the work starts on its own instead of in a panic.
• Train the team while there is time. Nobody learns a new tool well in the middle of the rush. Everyone can learn it in the quiet months. If the team is just you, that applies to you too.

The people who do this will not have an easy busy season. Nobody does. But they will have a different one, where the work that used to eat the last two weeks is already moving months earlier.

I did this in my own firm

This is not theory for me. I did exactly this in my own firm: used the quiet months to change how the work gets done, so the next season ran differently.

The result is something I had not had in 36 years in this profession. Work-life balance. Not a lighter season, an actual life outside of it, for the first time in my career.

I am not saying that to sound impressive. I am saying it because I spent most of those 36 years assuming the crunch was just the price of the job. It was not. It was the price of doing the job the same way every year.

What it looks like in practice

The free course is where to start, and it is self-paced, so it fits around whatever you have on right now and then into the quiet months:

• AI Skills Black Belt (free, for life): learn to use AI properly on the work you already do. This is the fall project.
• AI Agent Black Belt (Builder): build one agent around one real job, with quality checks. This is the winter project. For a firm, the document chase is a very good first agent. For a business or a job, pick the task you would hand off first if you could.
• AI Role OS Black Belt (Operating Systems): connect your people and agents into a weekly rhythm, so it runs without anyone remembering. This is what you have in place before the next busy stretch.

All three are updated as the technology changes, which matters when the tools you learn in October will have improved by spring.

The next two are $1,000 each. If you buy them in your first week after signing up for the free one, you get both for $1,000.

The cost of one tax return

Here is the easiest way I know to think about that price. Both courses together, in your first week, cost about what it costs to have one business tax return prepared.

If you run a firm, that is roughly the fee for one return. If you own a business, it is about what you pay someone to prepare yours. If you work a job, it is a single bill, once, for skills you will use every week for the rest of your career.

Now think about what that one return's worth buys if you spend it in the quiet months. The skills to build an agent that does one real job for you, over and over. A weekly rhythm that keeps it running without anyone remembering. If that saves you even a few hours a week, it has paid for itself many times over before April 15, 2027.

That is how you up your business game, or your career: not by working more hours next season, but by investing the cost of one tax return into how all the rest of your work gets done.

We would never tell a client, a customer or a friend to skip a $1,000 investment that pays back like that. Most of us are just not used to giving ourselves the same advice.

The point

The best time to change how you work is when it is already changing. If you are in the middle of a merger, a new hire, a new role or a software move, fold AI into it now. If you are not, the quiet months are your change, and for anyone in tax they start in thirteen days.

Nobody wants to change how they work, except during the three weeks when they have no choice. The off-season gives you three months. Use them.

AI does the typing. You make the judgment calls.

Where to start

• Free AI Skills Black Belt, self-paced, free for life: https://humansplusagents.ai/legacy-sbc
• Both next courses for $1,000 together, in your first week only: https://www.paypal.com/ncp/payment/KZ8E57JCPNC3U
• Claim a free AI agent to start on now: https://aiaccounting.legacysbc.com/claim-agent
• Join our free community, where people are planning their quiet months right now: https://www.facebook.com/groups/1024923770013348

 

What Happens After the Free Course

Every free course has a second page. You know the one. You finish the last lesson, feel briefly accomplished, and then a banner slides in with a countdown timer and a price that was not mentioned anywhere on the way in.

I have been on the receiving end of that page more times than I can count, and I have never once felt good about it. So this post is the second page, written before you start, with the prices in it.

The free one

Our AI Skills Black Belt is free. Not free for 14 days, not free with a card on file. Free for life, through my link, and nothing renews or expires.

It is nine belts on using AI for the work you already do: solving problems, learning faster, thinking through a decision, and creating the first draft of almost anything. It is self-paced, so you can do a belt on a Tuesday night and the next one three weeks later, and nobody sends you a sad email about it.

You keep it whatever you decide about anything else in this post.

https://humansplusagents.ai/legacy-sbc

The two after it

There are two more courses. They are where most people who finish the first one end up asking to go next, because the first one teaches you to use AI and these two teach you to make it do a job.

AI Agent Black Belt (Builder). You build one agent around one real job in your work, with quality checks built in, so you can trust what it hands you. For an accountant that might be the document chase, the month-end close checklist, or the bank reconciliation you have done forty times and could do in your sleep. That is exactly why it is the right one to hand off first.

AI Role OS Black Belt (Operating Systems). This is where the pieces become a week. You connect your priorities, your people and your agents into a daily rhythm, so the agent you built actually runs on Monday instead of living in a folder you mean to get to.

Both are self-paced, the same as the free one. Certificates are earned through work you actually do and that gets assessed, not handed out for watching videos.

And all three courses keep up. As the technology changes and the agents improve, the courses are updated, so what you learn this month is not out of date by spring. With AI, that is not a small thing. Half of what people learned about it last year now has an asterisk.

The prices, all of them

Here is the whole thing, in one place:

• AI Skills Black Belt: free, for life, and updated as the technology changes.
• AI Agent Black Belt (Builder): $1,000.
• AI Role OS Black Belt (Operating Systems): $1,000.

And the one offer, with its one condition:

• In your first week after you sign up for the free course, you can get both of the next two courses for $1,000 together. That is half price.
• After that first week, they go back to $1,000 each.

That is the entire offer. No countdown timer that resets when you refresh the page. No "only three seats left." The week starts when you sign up, not when I send an email, so it is your week, not mine.

https://www.paypal.com/ncp/payment/KZ8E57JCPNC3U

Why the first week

A fair question is why there is a deadline at all, if I am being so virtuous about fine print.

The honest answer is that the first week is when it works. People who move from the free course into building an agent while the first course is still fresh tend to finish. People who mean to come back to it later mostly do not, and I would rather give the discount to the version of you that is going to finish.

It is not a trick to rush you. If the first week passes and you decide you want the next two courses in March, they are $1,000 each in March, and they will still be worth it. You just will not get them for half.

What you do not have to decide today

You do not have to decide anything about the paid courses to start the free one. Start it, do a belt or two, and see whether it changes how your week goes. If it does, the next step is sitting right there with a price on it. If it does not, you have a free course and I have one more person who knows how to use AI properly, which is the point of all this anyway.

To be clear about where I stand: the free course on its own is a perfectly good place to start. But my recommendation is all three. The first one educates you. The second one improves one real part of your work. The third one is what makes it stick, by putting it into your job or your business every week. Learning AI without implementing it is a bit like buying the gym membership for the parking.

The best thing anyone has said to me about these courses came from a student who went through them:

"I was afraid of it, then I didn't understand it, and now I can do it all myself with real vision and understanding."

Afraid, then confused, then capable. That is the whole path, and it is the reason I recommend all three instead of stopping at the first.

AI does the typing. You make the judgment calls. That applies to this decision too.

Where to start

• Free AI Skills Black Belt, self-paced, free for life: https://humansplusagents.ai/legacy-sbc
• Both next courses for $1,000 together, in your first week only: https://www.paypal.com/ncp/payment/KZ8E57JCPNC3U
• Claim a free AI agent for your own work: https://aiaccounting.legacysbc.com/claim-agent
• Join our free community, where people post what they are building and answer each other: https://www.facebook.com/groups/1024923770013348

There is no upsell hiding in here. The upsell is right here, in the open, with a price on it.

 

Three Hundred Clients Still Owe You Documents, and October 15 Is Two Weeks Away

Yesterday morning I sat on a call with a tax firm we work with, and somewhere in the first ten minutes the owner said a number out loud: a little over three hundred extension clients who still owe documents.

Not three hundred returns that need preparing. Three hundred people who have not sent the thing the return is waiting on.

I had walked into that call with a picture in my head. I assumed the list lived in a client portal, with reminders going out on their own and a dashboard someone glanced at. It lived in an Excel export from the tax software, one row per client, and one person on the team was working down it by hand. My picture was wrong, and I suspect it is wrong about most firms, because it was the picture of how it is supposed to work.

The problem is not the preparing

If you do tax work, you already know this. The prep is the part you can control. Give a good preparer a complete file and the return gets done. What you cannot control is the file arriving.

Read your own extension list slowly and you will find the same people I found on theirs:

• The client waiting on a K-1 from a partnership that files on the last possible day, every year, on purpose.
• The one whose brokerage sent a corrected 1099 in August, so the first one is now wrong and the second one is in a drawer.
• The rental property owner who is "pulling together the expenses," a phrase that has meant the same thing since May.
• The daycare provider's tax ID, which nobody has, including the daycare provider.
• The client who will send everything tonight, and has said so every Thursday since June.

Every row has a status column, and every status column says some version of the same thing. Waiting on client. Since April.

What the chasing actually costs

Here is the part that gets missed. The person doing the chasing is usually good at their job, which is exactly why the list gets handed to them. And then their October disappears.

Each client needs a different email, because each one is missing something different. A 1040 client waiting on a K-1 needs a different message from an S corporation whose books have not been closed. The ones who sent half of what you asked for need to be thanked for the half before you ask for the rest, or they stop reading. Then you need to remember who you wrote to, when, and what came back. Three hundred times.

Nobody does that well from a spreadsheet. They do it until about the ninetieth row, and then the rest get the same email.

What we built for it

So we built the firm an agent to do the typing.

It reads the extension list straight from the tax software's export. It sorts every client by what is still missing and by how close their deadline is. It writes the chase email for each one, for every return type, in the firm's own voice: individual, S corporation, partnership, corporation, trust. It asks only for what has not arrived yet, and it thanks people for what they have already sent.

It works on a schedule you choose. At fourteen days out the email is friendly. At nine it is firmer. At six it names the deadline. At three the agent stops writing emails and puts the client on the phone list, because by then an email is a way of not having the conversation.

And it keeps the list current. As documents come in, the agent updates who has sent what and who is still waiting, so the person running the chase starts each morning with today's list, not last Tuesday's.

One thing I will say plainly: take the Social Security numbers out before anything goes near an AI tool. The agent does not need them to chase a document, and neither does anyone else.

AI does the typing. You make the judgment calls. The agent can tell you who is missing a K-1. It cannot tell you which client needs a phone call from you personally because they are going through something this year. That part is still yours, and it should be.

If your list looks like theirs

You have about two weeks until the October 15 deadline, which is enough time to change how the next two weeks go and not enough to keep doing it by hand.

You can claim a version of this for free. Ask for the Document Collection Chase Agent: https://aiaccounting.legacysbc.com/claim-agent

If you want to learn to build and adjust agents like this yourself, our nine-belt AI course is self-paced and free, and nothing renews or expires: https://humansplusagents.ai/legacy-sbc

And if you would rather compare notes with people whose extension lists look exactly like yours right now, our free community is full of them: https://www.facebook.com/groups/1024923770013348

The returns are not the hard part this month. The hard part is three hundred emails, each one a little different, sent by one person who also has a job. Give that part to the agent, and give the person their October back.

 

Your First Client Almost Always Already Knows You

Over the weekend a bookkeeper replied to a message I had sent her. She is qualified, she is ready, and she is having a hard time finding her first client.

She had found one. He turned out to be a scammer.

She also said the marketing company she had hired was starting to look suspicious. Then she asked me the question I get more than any other from people starting out: how did you get started, and do you have any tips?

I want to answer it properly, because the honest answer is a lot less exciting than the ones she has probably been sold.

First, the two things that are not clients

A real first client will never need you to send him money. Not to "cover a fee", not to pay a vendor on his behalf, not to return the extra from a check that came in too large. The overpayment is the scam. The check bounces a week after your refund clears.

Other signs worth knowing:

• He is always in a hurry and never wants a phone call.
• He cannot give you a business name you can look up.
• He asks for your bank details to "test a deposit".
• He wants you to buy gift cards, which no business in history has ever paid a bookkeeper in.

The client who needs you to send him money first is not a client.

The marketing company is a different animal, but the pattern rhymes. Be careful with anyone who guarantees a number of clients, wants payment up front for leads you cannot check, will not tell you where those leads come from, or locks you into a long contract before you have seen a single name. Anyone selling you guaranteed clients is selling to you, not finding clients for you.

None of that makes you naive for trying. It makes you normal. The people selling shortcuts to new practice owners are very good at finding new practice owners.

Where first clients actually come from

Almost every first client I have seen come in has come from someone who already knew the bookkeeper. Not from an ad. Not from a directory. From a person.

Think about who is already in your life:

• The former coworker who went out on her own last year and is doing her books at the kitchen table.
• The neighbor with the food truck who keeps his receipts in a shoebox, and says so.
• The hairdresser who does her own bookkeeping on Sunday nights and hates every minute of it.
• The guy from church who builds decks and has not reconciled a bank account since 2019.
• The person on your LinkedIn you have not spoken to in four years, who just posted that they started a business.

You already have a list. You have just never read it as a list of clients.

The part I got wrong, recently

The work is going through those people one at a time and saying something real to each of them. Not a post. Not "I am now taking clients!" to everyone at once. One person, one message, about them.

I know this works because it is how I spend part of most mornings. And I know how easy it is to get wrong, because I got it wrong last week.

Two people turned down something I had never offered them. One filed it as help she might ask for later. The other filed it as a feature she could add for her clients. It was neither. It was a free course for them, and my message had led with the offer instead of with the person. The fix was not a better offer. It was putting them first in the sentence.

That applies to you more than it ever will to me. When you write to the former coworker, the first line is about her new business, not about your services. The services come up because she asks.

Doing it without a spreadsheet of guilt

The hard part is not writing one message. It is writing forty over a few weeks without it turning into a list you feel bad about.

That is what our LinkedIn Connection Engine is for. It is the outreach system we run on our own list, packaged for yours. You export your LinkedIn connections, it sorts them into who to approach first, and it gives you ready-to-send messages with a space in each one for the thing that is actually true about that person. There is a follow-up sequence and a 30-day plan, because most replies come after the first message, not from it.

You can claim it free. Just ask for the LinkedIn Connection Engine: https://aiaccounting.legacysbc.com/claim-agent

And there is one more place worth being. Our free community is full of practice owners, and some of them hand out overflow work when they are busier than they can manage, which in accounting is most of the year. That is a very common way to land a first client and prove yourself on someone else's file before you have your own: https://www.facebook.com/groups/1024923770013348

If you want to build the skills while you build the list, our nine-belt AI course is free, self-paced, and nothing renews or expires: https://humansplusagents.ai/legacy-sbc

What I told her

I told her she was already doing the hard part, which is asking. Most people starting out do not ask anyone anything. They buy the thing that promises they will not have to.

Your first client probably already has your phone number. They just do not know you are open for business yet.

 

Stop Asking What It Is Worth. Ask What You Could Do With It.

I was writing an email on Thursday and I typed a sentence I have now typed something like forty times this month. Normally a thousand dollars per person. Free right now, nothing renews, nothing expires.

I read it back and thought, that is a good sentence. It is clear, it is true, and it answers the first question anybody asks.

Then I watched what people did with it, and I changed my mind.

Here is what I got wrong. The moment you tell somebody the price of a free thing, you have turned it into a bargain. And a bargain is not something people do. A bargain is something people save. It goes in the drawer with the other good deals, you feel briefly clever about it, and then it is March.

I know this because I do it. I have a folder of free things I fully intend to get to. It is the best organized folder I own.

So I would like to stop talking about what it costs, because the cost was never the interesting part.

The question that actually decides it

The useful question is not what is this worth.

It is this. What would I be able to do in three weeks that I cannot do today?

That is a different question and it has a different kind of answer. Price is a claim somebody else makes about a thing. Capability is something you can test on a Tuesday afternoon with a real client file and find out for yourself.

What is actually in the arsenal

Nine belts, self paced, twenty seven worked examples. But belts and modules are not skills, so here is what comes out the other end, in the language of the actual job.

You can look at something you do every single month and tell whether it should be a checklist, a prompt or an agent. Those are three different things and almost everybody treats them as one thing.

You can write the prompt that turns a month of messy transactions into a categorized draft. More to the point, you can tell when the draft is wrong, which is the skill nobody advertises.

You can build one agent around one real job with a quality check inside it, so that when it fails it fails loudly. Quiet failure is the entire risk. A tool that is wrong and confident is worse than no tool.

You can explain to a client, out loud and without hedging, why the machine's answer does not fit their situation. That has always been the billable skill. It was billable before any of this existed.

And you can sit in a meeting where somebody says we should be using AI for this, and be the person in the room who can say what that would actually involve, on Monday, with the staff you have.

Read that list again slowly and notice that none of it is about typing faster.

AI does the typing. You make the judgment calls. The course is about the second half.

The part that is less comfortable

There is a second reason to do this and it has nothing to do with your own workflow.

Businesses are paying for implementation help right now. Not for a talk about AI. For somebody to come in, set the thing up properly, and then stay long enough to be sure it has not quietly broken anything.

Almost nobody can do that work yet. Not because it is hard. Because the people who understand the tools have never closed a month in their lives, and the people who have closed a month have not been given a week to learn the tools.

You are already holding the half that takes ten years. The other half is the part being handed out.

What I am doing differently

I am going to stop leading with the price.

The honest version is this. It takes about an hour a day. Nothing renews, nothing expires, there is no card, and if you stop halfway through, nobody emails you about it.

That last part is genuinely a feature and it is also the danger, which is why I would put a day in the calendar rather than filing it somewhere safe. You already know which folder it would go in.

Start here if you are a bookkeeper or an accountant: https://humansplusagents.ai/legacy-sbc

Start here if you are anything else, and plenty of the people reading this are: https://humansplusagents.ai/legacy-sbc-aia

If you would rather try something smaller first, claim a free agent and point it at a real file this week: https://aiaccounting.legacysbc.com/claim-agent

The community is here, and it is free too: https://www.facebook.com/groups/1024923770013348

One last thing, and I am deliberately putting it last. If you own a practice, you can give this same course to every one of your clients, free, for good, with no cap on the number. I have been leading with that for a fortnight and I think it has been crowding out the thing that matters more, which is what you would be able to do yourself by the end of the month.

Five times this week I told my AI it was wrong

 Five times this week I told my AI it was wrong

I sat down Friday morning with a report my AI had written about my own business, and the first thing on it was a problem that did not exist.

It told me one of our products was on sale for twenty-seven things when only seven of them had been built. That is the kind of sentence that makes your stomach drop, because if it were true it would be the worst thing on the page and it would be my name on it.

It was not true. The seven were the new ones. The other twenty already existed and had been pulled in from elsewhere, which is how every single one of these products is assembled and has been for months. It took me one line to say so. It had taken the machine a paragraph, a table and a confident heading to be wrong.

I want to be honest about my first reaction, because it was not clever. My first reaction was to believe it. It was well written. It had specifics in it. It cited a file. Everything about the shape of it said this has been checked.

That is the whole problem, and it is why I am writing this instead of something more useful.

The other four

Once I started counting I got to five before lunch, and they were all from this week.

One. It went looking for how many of my contacts work in the car repair trade, to decide whether that was a market worth building for. It came back with six. The real number was one. The other five were the letters t i r e sitting inside the word entire, and inside the word retirement. I had a morning of quiet enthusiasm about a market that turned out to be financial advisors.

Two. Same exercise, different trade. It reported thirty-seven contacts in manufacturing, which would have made it the biggest opportunity on the list. The real number was four. The rest were a business coach who works with manufacturing leaders, somebody in manufactured housing, and, my favourite, a note the machine had written itself a week earlier saying you cannot quickly manufacture reviewers. It counted its own sentence as evidence of a market.

Three. It needed the current web address for our community. There are two, an old one and a new one, and the old one was retired three weeks ago. It searched every file we have, found the old address two hundred and eighty times and the new one far less often, and concluded the popular one was the right one. It had run a poll. The documents all voted. The documents were all from August.

Four. It needed the next available number to file something under. A tracking document said number two hundred and forty-nine was free. Two hundred and forty-nine was not free. Something had been filed there days earlier and the tracking document had not been updated, which is the entire reason you do not trust a tracking document.

Five was the one I opened with.

Read that list again and notice what is not in it

Not one of those needed technical knowledge to catch.

You do not need to know anything about how these systems work to know that entire contains tire. You do not need to understand anything at all to think, hang on, thirty-seven seems like a lot. The question that caught the last one was five words long and it was not a clever five words. It was that is not how those work.

Every correction this week came from somebody who knew the business, not from somebody who knew the technology. That is not a small point. It is the whole point, and it is the opposite of what most people assume when they sit down in front of one of these things for the first time.

What I got wrong for about a year

I used to think that checking AI output meant being able to out-think it. That to say no, that is wrong I needed to already know the right answer, and if I did not already know the right answer then who was I to argue.

That is completely backwards and it kept me quiet for longer than I would like to admit.

You do not need the right answer. You need to notice that something is off. Those are different skills and the second one you already have, because you have been doing it to junior staff, to vendors, to bank reconciliations and to your own numbers at eleven at night for your entire career. The feeling you get when a set of books looks fine and something is still wrong, before you can say what, is exactly the muscle this needs. Nobody told you that was transferable. It is.

Why you do not push back

Here is the part I think most people will recognise and nobody says out loud.

It is uncomfortable to argue with something that sounds certain. Especially when you are new to it and it is not. The output arrives fluent, formatted, complete, with headings and a summary and a confident tone, and pushing back on it feels a bit like putting your hand up in a meeting to question the one person who clearly knows what they are talking about.

That is a confidence problem wearing a technical problem's coat. And it gets worse the less experienced you are, which is exactly backwards from where it should sit, because the people who most need to challenge it are the ones who feel least entitled to.

The three questions

I do not have a clever system for this. I have three questions I ask every time now, and the first one does most of the work.

Where did that number come from? Not is it right. Where did it come from. Made it recalculate the manufacturing count and cut it from thirty-seven to four.

What would make this wrong? Asked of anything confident. It forces the thing to argue against itself, and the good ones will.

Show me the thing, not the summary of the thing. The summary is where the errors go to hide, because a summary is polished by definition. Four of this week's five survived right up until somebody asked to see the underlying list.

One more thing

I am not telling you this to make anybody nervous about using these tools. I use them all day and they did an enormous amount of good work this week that I have not mentioned once, because correct work is not interesting to write about.

I am telling you because there is a version of this where somebody reads a confident paragraph, does not want to look slow by questioning it, and lets it through. That is the only failure mode that actually matters, and it is not a technology failure. It is a nerve failure.

So question it. It will not be offended. It is, if anything, slightly better than most people at being told it is wrong, which is the single most underrated thing about the whole business.

 

Who trains the next reviewer?

I had two conversations this week that I could not stop thinking about, and the annoying part is that neither person was trying to make a point. They were both just describing their own Tuesday.

The first was a senior accountant at a firm in Brooklyn. Senior accountant is a strange promotion. Nobody throws a party. What actually happens is that the difficult files start appearing on your desk, and one day you look up and realise you have stopped being the person who prepares the work and started being the person who checks somebody else's.

Those are not the same job. Preparing is procedural. Checking is diagnostic. You look at a finished reconciliation and you know, in about four seconds, roughly where the problem is going to be hiding. Nobody teaches you that. It shows up, or it does not.

The second was a man who runs offshore accounting teams for Western firms. He had been thinking about what happens when the machine assembles the file and a person signs it, and he said it better than I have managed to:

"The valuable version of my team is not the one that does reconciliation, it is the one trained to audit a prepared reconciliation fast and catch what is wrong in it. That is a different hiring and training problem than the one I have been solving for. Pods built for production do not automatically become pods built for verification."

He is right. And I have been sitting with it for two days because there is a second half he did not say and I did not think of until later.

The problem is not that the two skills are different.

The problem is how the second one has always been made.

---

You already know this. You have just never had to say it out loud.

Think about how you actually got good at review.

You did not read about it. There was no session on it. You built several hundred reconciliations, a healthy number of them were wrong, and somebody more senior sent one back with a single question and no other comment, which is somehow worse than being shouted at. And then you did it again. And somewhere in year three or four you started seeing the error before you found it.

That instinct is a residue. It is what is left over after enough repetitions of the boring work, done badly, then corrected in public.

The production years were not a toll you paid on the way to judgment. They were the machine that produced it.

Which brings me to the question I have not been able to answer.

If the machine does the first four hundred reconciliations, where does the fifth-year reviewer come from?

Your current seniors are fine. They already have it, they earned it the old way, and nobody can take it off them. The people who would have become your seniors in six years are the ones I would worry about, because they will not have had the six years. They will have had six years of checking work that was mostly right, which teaches you almost nothing, because the thing that teaches you is being wrong in front of somebody who noticed.

---

The comfortable answer is wrong and I want to say so before someone else does

The comfortable answer is that we keep juniors doing manual work in order to season them, like a cast iron pan.

That is a bad answer. It is also never going to happen. No firm is going to pay for deliberately slower work as a character-building exercise, and no junior is going to sit quietly through three years of data entry while the person at the next desk finishes at four.

The real answer is less tidy, and it is this. If the training used to be a by-product, it now has to be on purpose.

Somebody has to design it. Not buy a tool. Design a way for people to build the judgment that the grunt work used to hand them for free, in a world where the grunt work has left the building.

I do not have a finished answer to that, and I would be a bit suspicious of anyone who says they do eighteen months in. What I do have is the part that comes first, which is knowing what you are actually training.

---

The good news, and it is better than I expected

Here is where I changed my own mind this week.

Reviewing a junior and reviewing an agent turn out to be the same skill.

Both hand you work that looks finished. Both are confident. AI is wrong the way a very polite person is wrong: fluently, in a complete sentence, with excellent formatting. And both are right most of the time, which is exactly what makes the occasional miss dangerous, because you stop looking properly.

In both cases the question you are asking is identical. Not "is this wrong" but "where would wrong hide in this."

So if you have ever reviewed another person's work, you already have the beginning of it. You are much further along than you think, and considerably further along than somebody who is brilliant with the tools and has never once had to sign their name underneath numbers they did not prepare.

That is the whole reason our training is built around judgment rather than tooling. Not because the tooling is unimportant, but because the tooling changes every few months and the judgment does not. AI does the typing. You make the judgment calls.

And it is why the hardest belt is not the one where you build something clever. It is the one where you decide what not to automate, and where to stop trusting what comes back.

---

The same problem, from two different chairs

If you work on the books, this is a career question, and the honest version is uncomfortable. The part of your work that is easiest to hand over is the part you are currently fastest at. The part that is hardest to hand over is the part you probably undersell, because it feels like something anybody could do. It is not. I promise you it is not.

If you run a company, this is not a career question at all. It is a workforce design problem on a very long fuse.

You will get the productivity. That part is real and it shows up fast. What shows up slowly, and only if somebody builds it deliberately, is the bench underneath your current senior people. Nobody notices a broken training pipeline in year one. You notice it in year five, when the person who should have been ready is not, and you cannot make up the ground quickly because the ground was made of repetitions that do not exist any more.

The firms that come through this well will not be the ones with the best tools. They will be the ones who worked out early how to make reviewers on purpose.

---

Both doors are open and both are free

If you work on books, the nine-belt AI Native Foundations course is free for life. It normally costs $1,000 per person. Twenty-seven worked examples built on the real work rather than a case study, and the belts that matter most for everything above are the ones about judgment and knowing when to stop trusting the output.

And you can gift that same course, free for life, to every client you have. Uncapped. Not a trial and not a discount code. Your clients are one rung down the same ladder, and it is a reason to be in touch that is not an invoice.

https://humansplusagents.ai/legacy-sbc

If you run a company with ten or more employees, the entire first year of the $25,000 programme is free. All nine courses, a complete AI-Native Company Transformation Plan, monthly CEO peer-group forums, and Foundations training for every employee you have.

That last part is the one that matters here. Training everybody in the building is how you rebuild the bench on purpose rather than hoping it happens by accident. Ten employees is a real threshold rather than a suggestion, and year two is $25,000. Sign up free first, then request CEO access from the offer bar inside your account.

https://humansplusagents.ai/legacy-sbc-aia

Nothing expires. No card. Nothing renews.

---

We are all working this out at the same time, and the people who will figure it out first are the ones already doing the work rather than reading about it. If you have a view on where the next generation of reviewers comes from, I would honestly like to hear it. I am still short of an answer.

The community is free and this is mostly what we talk about: https://www.facebook.com/groups/1024923770013348

And if you would rather start with something small, we build free agents one job at a time, yours to keep, no course required first: https://aiaccounting.legacysbc.com/claim-agent