letter.md
Your Business, Written Down
I want you to remember a specific week. Maybe it was eight years ago, maybe twelve. The business was small, the stakes were real, and nearly every hour you worked was the work itself. You coached, you taught, you built the thing, you sat with the client. When someone at a dinner asked what you did, the answer was one sentence, and it was true.
Then it worked. The business grew, and growth, up close, is mostly made of decisions. Someone has to decide what happens when a payment fails twice, whether the refund window bends for the client with a good story, what to say when a member asks to move her session a third time. In a founder-led business those decisions all learn the same route. They come to you. Each one made sense on the day you took it. String ten years of those days together and you are running a company whose real operating manual exists in one copy, in your head, and the only way anyone on your team can read it is to interrupt you.
You know the costs better than I do, so I will not repeat them at you. The questions that follow you onto vacation. The team that waits, because asking you is safer than guessing wrong. The strange arithmetic where you hired four people and somehow kept less of your week. And underneath all of it the quiet one, that the work you started the business to do now gets whatever attention is left over, and everyone can tell, including you.
Here is my wish for you, and it is the same wish I bring to every founder I work with. I want the business to know how to run itself the way you know how to run it. I want your team finding answers without finding you first. And I want your best hours back on the work only you can do, because that work is why the business exists, and no software is ever going to do it for you.
There is one way I know to get there that actually holds.
The programmable organization
The version I build with founders starts with one move. We write the business down. The rules you have been carrying, every one of them, come out of your head and out of the settings screens of your tools and into plain English files in one folder. What happens when a payment fails a third time. How a refund gets decided on day 31, and the exact voice the answer goes out in, because it goes out in yours. What a new client's first week contains, every email, every access grant, every check that each one happened. Which three numbers you actually watch, and what the team does the week one of them dips.
Then AI reads the folder and does what it says. It answers the reschedule email the way you would have answered it, files what came in overnight, drafts the Thursday send, runs the checks, and brings you the two things that genuinely need you. You do not log into it the way you log into your platforms. It works for you, and the folder is its employee handbook.
Better models keep arriving whether you ask for them or not. When one ships, you hand the same folder to a better worker. Nothing gets rebuilt, because the business was never inside the tool. It was in the files, and the files are yours.
Not everything in the folder needs a mind, either. A surprising share of your operations are pure rules, the receipt, the retry, the reminder, the weekly numbers, work that should run the same way every time. Rules can be code. So alongside the worker, the folder governs a handful of small, boring programs you own, and the worker keeps them repaired. The judgment stays with you, the rules stay written down, and the machinery underneath stops being anyone's mystery.
A folder like this ends up holding more than procedures. It holds who you are, the things your business always does and the things it never does, so the worker inherits your judgment instead of a vendor's defaults. The answers you have given a hundred times go in, written once. What you are trying to make happen this quarter goes in, so the daily work has something to point at. So does what is happening right now, the one part your dashboards already had. For the first time, everything the business believes lives in a place your team can read.
A business run this way has a name: a programmable organization. Lee Bryant argues for it at enterprise scale, where a change program and a transformation budget carry the shock. I run the version for companies where the entire change program is you. It turns out the small version is the more personal problem, not the smaller one.
And the folder, for all the weight I just put on it, is one lever among several. Swapping the worker when a better model ships is a second. Owning the small programs instead of renting them one subscription at a time is a third. Emptying the stack a row at a time is a fourth. And the people are the last and largest, you first, because nothing above holds unless the humans change with it.
Where automation left you
Before you file this under automation, I want to name what you already own, because you have been sold automation for fifteen years and you have the stack to show for it. There are dozens of small machines in there, maybe hundreds, spread across a dozen subscriptions, maybe two dozen. Your launches depend on some of them. Nobody has opened others in two years. One business I audited was running more automations than anyone on the team could count, and the map we made was the first time anyone had seen them in one place.
And the stack bills you for the privilege. Somewhere in your books is a subscription line that adds a row or two every year and never loses one, the scheduler, the course platform, the email machine, the forms, the connector that exists only because two of your tools refuse to talk to each other. Each row holds a slice of your business, priced so that leaving hurts more than staying. When one tool changes its pricing, or quietly breaks the thing another tool depended on, you find out at launch. Keeping the duct tape holding is a job in itself now, and it is usually yours, because only you know where the tape is.
Automation, the way you have lived it, meant moving your rules into the tools, one settings screen at a time, in a language only the tool speaks. Every trigger and campaign and zap is a decision you made once, frozen somewhere nobody can read it, least of all you. So the stack grew while the understanding shrank, and the gap between the two became your job. You are not the bottleneck because you failed to automate. You are the bottleneck because you automated the way the tools wanted, and the tools never wanted your rules anywhere your team could read them.
Here is what changed underneath all of that. Renting made sense when owning was impossible, when custom software meant one developer to build it and another to keep it alive. That arithmetic died quietly, and not just because code got cheap to write. Code got cheap to keep. The same AI that reads your folder repairs the small programs the folder governs, which means a business your size can now own the rule-shaped half of its operations for less than it pays to rent it. On the work that is just rules, familiar is the only advantage the stack has left.
Writing the business down reverses the direction of travel, and then it starts emptying the stack. The rules come out of the tools and into files a person can read, and the tools go back to being hands. They were never meant to be the memory. From there the migration is one unglamorous move at a time. The audit shows which subscriptions were only ever a rule with a login, and those become a page in the folder and a small program you own, and a row leaves the bill. The tools that genuinely earn their keep stay, holding work instead of holding rules. Nobody moves their whole business in one weekend, least of all with me in the room. The day the rules have an address, your team stops needing yours, and once the boring work is code you own, the bill finally stops growing.
The checking phase
Founders ask what the first month feels like, and the true answer is that it feels like checking. The first time the week's numbers arrive without anyone pulling them, you read every line against the version in your head. The first time the worker answers a reschedule request, you open the sent message and read it twice, the way you would read over a new assistant's shoulder. You should. I would never ask a founder to trust a system on principle, and I have watched trust arrive on schedule anyway, because it arrives here the way it arrives with people, one checkable detail at a time. I run my own company from a folder like this, so I know the checking phase from the inside.
Then a particular morning happens, and it is not dramatic. The numbers are right, the two flags raised are the right two flags, a client got her new session time in your voice while you slept, and you notice you did not check anything yesterday. Founders report the same two feelings in the same breath. Relief first. Then something quieter and stranger underneath it, because the business needing you every hour was also, for years, the daily proof that you mattered. Watching it need you differently takes getting used to. I bring that up in the first conversation, every time, because pretending the feeling is not coming does not stop it coming.
The team crosses its own line around the same time. Questions that used to wait in your inbox start going to the folder, and because the folder answers in writing, the answer is the same on Thursday as it was on Monday. New people learn the business by reading it instead of by interrupting the two people who remember. The operating manual finally has a second copy. The one in your head stops being load-bearing, which is a strange thing to want for yourself, and every founder I have done this with has come to want it within the month.
What the folder cannot do
Here is the part I refuse to sell past. The writing down is not the hard part. My tooling reads what is already sitting in your stack faster than you would believe, and the first draft of the folder lands in weeks, not quarters. The hard part is that a business cannot be written down without being decided, and deciding is owner's work. Half the rules in your head are really two rules that contradict each other, held in balance by your judgment on the day. The refund policy is whatever the page says, except when it is what your gut says. Writing that down means the business finally has to mean something by its own rules, and you are the only person who can say what it means.
And then it starts working. The folder answers well, again and again, and what sounds like the finish line turns out to be the start of the real change. You built a company where safety meant asking you, and your team learned that thoroughly because you taught it thoroughly, one small approval at a time, for years. Now safety has to mean the folder, and asking has to become the exception that improves a rule rather than the habit that routes around it. That is a change in what people believe, and software does not install beliefs. Someone works that change with you and your team, week after week, while the new habit takes. That someone is me, and it is why I say I coach the change instead of delivering it.
The engineering side keeps getting easier. The models read a folder more reliably every few months and do more of what it says. What does not get easier on any schedule is the person at the center of the company unlearning the job of being its memory. I have watched software eat every other part of this work. It has never once done that part.
I know this turn from a second angle. For more than a decade I have coached operators and executives who reached the top of a game and felt the question underneath it change. The founder inside a programmable organization makes a version of the same turn, from being the answer to deciding what the answers are. It is one conviction with two doors. Manor points it at an organization. My coaching practice points it at one person at the turn, referred one conversation at a time.
A place to start
My wish for you has not moved since the first paragraph. I still want the business to know what you know, your team finding answers that do not route through you, and your mornings back on the work only you can do.
The way I start is deliberately small. Name the thing in your business that cannot happen without you. The report only you can pull, the decision only you can make, the question that always finds you at nine at night. Write me two paragraphs about it, and I will map what is standing behind it, which systems it touches, which rules it depends on, where those rules currently live, and send the map back. If the sharper pain is the bill, send me the subscription line from your books instead, and I will mark which rows are just rules you could own. Free either way, because it is the fastest honest way for both of us to see whether this work fits your business. If the map shows you something, the next room is the audit conversation, where I read your whole stack the same way, read-only, and you see the entire business at once, maybe for the first time since it grew past you.
On proof, I keep it modest until the people in it agree to be named. A recent audit surfaced five figures of leaked revenue in week one and mapped the full customer journey. The founders behind sentences like that one decide when their names appear, and that discipline is part of what you would be buying.
If you arrived here as a consultant instead, sitting on AI-era ideas of your own to judge, this letter has a companion, The Service Is the Moat, and it ends with an offer built for you.
The business you wanted is not gone. It is underneath the one you have, written in a language only you currently speak. Let's write it down.