Everything Point does on your behalf is written down. Which message Point touched, and what happened to it. When it happened, and on what reading. It’s set out in plain words rather than in a code, and the thread itself sits a click away from the row. Most of those actions can be put back.
That’s the third of the six questions answered. The answer is duller than the question sounds, which is usually a good sign. What deserves the rest of a chapter is why the record matters, because the obvious reason is the smaller one. Catching the software out is what you’ll open the record for perhaps three times in a year. Its real work is quieter, and it runs the whole time. The way to see it is to look at how you already come to trust work you didn’t watch being done.
Why a record beats watching
The reflex when you hand something over is to supervise. Watch it for two weeks, satisfy yourself, then relax. The instinct is sound. The instrument it reaches for is the wrong one here, and your own firm has always run on a different one.
Think about how work you didn’t see gets accepted in a practice. Somebody does a piece of it while you’re with a client. The whole thing happens out of your sight, and you don’t do the work again to check it. What you look at is what they left behind. The figures, and where each one came from. What was assumed, and what was queried by whom. The work becomes trustworthy because somebody who was out of the room can reconstruct it afterwards. Every trade that has to answer for work carried out of sight has arrived at some version of that arrangement. Watching is the one they all passed over.
That’s the instrument the first chapter said this was going to need. A model’s prose gives your eye nothing to catch. A summary that’s satisfying to read tells you how it reads, and that’s all it tells you. A record works the other way. It’s made of events. Each one has a time, a named thread, and a place in the order. Each one either happened or it didn’t. A paragraph is something you judge. An action is something you look up.
A wrong outcome you cannot place
Here’s where a record earns its keep, and the scene is more ordinary than the one people picture.
Four weeks in, on a Tuesday, something is off. A thread you needed turns out to have been filed. Or a client who answered you in March has just been chased for the thing they answered about. The relationship survives it easily. It’s still enough to make you sit back.
Take that Tuesday with nothing to consult. You know the arrangement produced a wrong outcome, and what produced it stays open. A one-off. A setting sitting a notch too high. Something Point picked up in week two, from a message you happened to deal with strangely. Your own hand on a Friday you barely remember. A fault with no location puts the whole arrangement under suspicion, and that’s the proportionate response while the part that failed stays out of reach. That’s what happens to people. One unexplained Tuesday ends most trials of most tools of this kind. The owner walks away holding a verdict about an entire category, assembled out of a single event they never got to the bottom of.
Take the same Tuesday with the record open, and the event has coordinates. Filed at 8:40, on the reading that the thread was a receipt rather than a client. It stops being a mood about software and becomes a fault in one place. A fault with a location can be repaired. One setting moves, or one learned habit gets corrected. Everything else stands, because the record put the fault somewhere.
That’s the defect the first chapter found in vague caution, met again from the other end. A doubt with no parts in it stays where it is forever. Evidence with no parts in it leaves it there. What the record protects is your ability to revise your view of this software. While you’re still early, a view you can revise is the only kind worth holding.
What a way back is worth
The other half of the answer is that most of what appears in the record can be put back. The value of that sits somewhere other than where people put it.
The value is price. Being wrong inside your own account costs a click, so you can find things out by trying them. That’s how you learned every other piece of software you rely on. The manual for the filing portal went unread. You pressed something, watched what it did, and pressed it back.
A decision that stands for good has to be made on faith, in advance, out of a supplier’s description of itself. A careful person withholds faith on those terms, and is right to. That’s exactly why careful people spend two years on a question that two weeks of use would have closed. A way back turns the whole thing from a decision you have to reach into an experiment you can run.
What comes back is what Point did inside your own account, and that’s a wide territory. A thread Point filed returns to the feed. A task Point lifted out of a message can go. A draft Point prepared can be discarded before anybody else sees it. The ranking, the labels and the tidying done while you were out all sit in a place you administer. Standing instructions work the same way. Ask to be told when a particular client replies, or when a deadline shifts. That watch is listed where you can see it, and it stops whenever you decide. That closes something the outbound perimeter left open, because supervising an instruction you gave in March means seeing in November what it has done since. All of it stops at the same edge as before. A message sitting on somebody else’s server is past every mechanism described here.
The expectation you do not have yet
You have something about a colleague of four years that you’ve never had to put into words. You know what they’ll get right unasked. You know where they go thin, and which kind of thing they’ll bring to you as a question. It’s worth more than any procedure you’ve ever written down, and you didn’t get it from their finished work. You got it by watching a few hundred of their small decisions go past, in order, over months.
On the first morning you have none of that about software. The output is the wrong place to look for it, for the reason running underneath this whole series. A sound decision and a poor one arrive in the same untroubled prose. The information lives in the sequence, and only there. Twenty actions in a row tell you what Point reached for first, and what Point treated as routine. They show you where a reading of a thread ran thinner than yours would have. Twenty finished summaries tell you almost nothing.
So the early reading is how the expectation gets built, whatever it looks like from outside. Do it properly, because it has an end. Once you have the sense of the thing, the record stops being a study. It becomes something you open on the rare morning when something surprises you. Reading it closely for a while is how you earn the right to stop. What a row has to contain before it’s worth reading at all is worked through in the activity log itself. So is how to read one without turning it into a second inbox.
Being able to answer for it
One more use, and it’s the one that counts on the day it counts.
A client calls about something that went out under your name. Why was I chased for a document I sent you in March. Two answers are available to you, and only one of them is survivable. The first is that the software did it and you’ll look into it. That tells a client something worse than the original error. It tells them things happen inside your practice that you can’t account for. The second answer is specific. This is what happened. This is when. This is the reason it happened. This is what has changed since.
Every firm produces errors, and clients have always known it. What they weigh is whether you knew, how quickly you knew, and what you did about it. Responsibility stays where it was, and a record leaves it there. The work went out under your name, and you’re the one holding the phone. What the record changes is whether you can discharge that responsibility or only apologize for it. What you owe a client in writing about any of this is a separate question. So is whether the arrangement belongs in the terms you send them at all. Both are answered in what to put in your engagement letter about AI.
What the record is really for
Every row is written after the event. Nothing here prevents anything. That’s worth saying flatly, because a thorough record is comforting, and the comfort will try to do work the record can’t.
It’s a division of labor, and seeing the division is the point of this chapter. Prevention is a setting. The record is the instrument that tells you where the setting ought to sit. Without one, how much a piece of software does on its own is a single act of faith. You place it once, in advance, and afterwards you learn nothing about whether it was well placed. That’s why the choice tends to collapse into all or nothing, and why it lands on nothing. With one, the move gets much smaller. Raise a single kind of work. Read what Point did with it for two weeks. Leave it there, or put it back where it was.
The three answers so far have each been a different shape. Reading was broad, and you can withdraw it. Sending was narrow, sorted by weight, and permanent at the far edge. The record is complete about what was done and silent about what nobody did. That silence is the absence that gives off no signal, and it sits past what a log could fix. All three are conditions that make a decision possible. The decision itself is the next thing.
The decision is the dial. It’s held separately for each kind of work, it sits on review when Point arrives, and it points at everything a mailbox can be handed. It’s the only thing in this series you move yourself, and it’s the question the whole argument has been walking toward.
So: how far, on what, and what it costs when a setting sits one notch higher than it should have.