You asked to be told about five things this year, and you can name two of them. That’s the ordinary life of a standing instruction. It works, so you stop thinking about it, and then one Tuesday something tells you about a matter that closed in June. Every watch Point is holding for you sits in one list you already read. Each row opens into the full record of what it has caught. Any of them ends in a click.
- Every watch is a task assigned to Point, in the same list as your own work. That’s why the list actually gets reread.
- Open a row and the record is what makes a standing instruction answerable. It shows what the watch looks for and whether it recurs. It counts the fires, and every fire routes back to the message responsible.
- A watch that has never fired is the ambiguous one. Nothing happened, or it’s aimed at the wrong thing, and from the outside those look identical until you check once.
- You have two levers. Sometimes the watch is right and the way it reaches you is wrong, and those are two different switches.
The list is one you already read
Most people go looking in preferences, find notification toggles and a signature, and conclude they have nothing running. Your watches live somewhere better.
A watch is a task. It sits in your task list, assigned to Point rather than to you and marked as watching, alongside the work you set yourself. What you asked for three months ago is a line of plain English in a place you open every day, and that’s the whole reason this arrangement is worth anything. A settings page is somewhere you visit once something has already gone wrong, so a stale instruction can sit in one for a year. A task list is somewhere you already are.
Everything standing is in that one list. The watch on a message you’re waiting for. The watch that carries a next step as well as a trigger. The scheduled check that goes out to the web on its own timer and never touches your mail at all, which is watching things outside your inbox and lives here with the rest of them. One place, one kind of row, one inventory to keep.
If you came to this page because something is pestering you right now, start with the notification rather than the list. The entry behind the bell names which watch fired and when. One click from it opens the message that set it off, so you go from the thing annoying you to the sentence that caused it in one step. That’s the fastest route to the row you actually want, and on a busy list it’s the only route worth taking.
One thing to expect when you get there. Most of what’s in that list is your own work rather than a watch. Point lifts asks out of the mail people send you and dates them without being told, so the list has been keeping itself since the day you connected, and email task management, without writing it down is where those came from. Your watches are the rows assigned to Point and marked as watching. Everything else is work.
What the record behind a row proves
Open one and you get the whole of it. It takes about fifteen seconds to read.
The instruction, in the words you used. Your own sentence, kept as you wrote it, which is the only version you have any hope of judging three months later.
Whether it ends or stands. A watch on the signed letter from Alvarez is done when the letter arrives. A watch on failed backup notices from a client’s server stands until you stop it. The record says which one you have, so you can see at a glance that something is still live.
How many times it has fired, and when it last did. This is the most useful number on the page and the one to read first.
And under that, the fires themselves, each one linked to the message that caused it. That link is what turns a standing instruction into something you can settle. You can follow any watch down to the exact conversation it caught.
Which gives you a test that takes two minutes and answers the only question worth asking about a watch you have forgotten. Open its last three fires and read the messages. Would you have wanted to be interrupted for those three? If the answer is yes three times, leave it alone and stop thinking about it. If two of the three are things you didn’t care about, the description is catching too much, and the fix is a narrower sentence rather than a stricter habit. And if every fire is the same automated notice from the same system, arriving on a schedule somebody else set, you have got the problem in turning ten alerts into one.
Reading the catches beats reasoning about the wording. A sentence you wrote in June always sounds right in September. What it actually caught is a fact.
A watch that has never fired
Zero is the reading that stops people, and it should, because zero means two opposite things.
It can mean nothing has happened. The letter hasn’t come back, the board hasn’t written, the season hasn’t started. The watch is doing exactly what it was set up to do, and its silence is the product working.
Or it can mean the watch is aimed at something it will never catch, while you have been running on the belief that you’d be told. That’s the expensive version, and it stays quiet about itself. Both look like a quiet row in a list.
Settling it takes one search. Go and look in your own mail for the thing that should have set it off. Three things can come back.
Nothing. The event hasn’t happened. The watch is fine and there’s nothing to do.
It happened, and the watch stayed quiet. Now you know something worth knowing. Almost always the sentence was written around words rather than around substance, so it was hunting for a subject line the other side never wrote. Rewrite it wider and set it again. How to phrase one so this stops happening is the subject of telling your inbox to notify you when something arrives, and it’s worth ten minutes if you have more than one of these.
It happened, you found out another way, and it was fine. This is the quiet one nobody expects. Your bookkeeper mentioned it, or you saw it near the top of your feed on the morning it landed, and nothing was lost. That watch works. It’s unnecessary, which is a different finding with the same answer: end it. A watch that has been carrying no weight is still costing you a row and a small amount of belief.
There is a fourth case that needs no search, and it’s the most common one on a list nobody has read in a while. The reason has gone. The matter closed, the client left, the server was replaced in the spring. The row looks right, because a watch pointed at something that no longer exists behaves precisely like a watch pointed at something that hasn’t happened yet.
Scheduled checks against the outside world go quiet in their own particular way. A check runs and reads every morning while it says nothing, and a page that has been returning an error since March produces exactly the same silence. That has its own confirming habit and it’s set out next door.
Quieting one without ending it
Before you end anything, be exact about what’s bothering you, because two different complaints get spoken the same way.
One is that you’d rather not know. The watch is catching things you’d have left alone, or the reason it exists has gone. That’s a watch problem and the answer is in the next section.
The other is that you do want to know, but not like this. The information is right and the delivery is wrong: a phone lighting up at 6:40 on a Sunday, a copy landing in your mail as well as the bell, a chime in a client meeting. Ending the watch there costs you the thing you actually wanted.
How Point reaches you is a separate set of switches, one each for email copies, the daily briefing, sound, vibration and system banners. Silence is a toggle. Turn the sound off and the watch still works. Turn the email copies off and the bell still holds the receipt.
Here’s the one worth pinning down, because it’s the mistake people make in the moment. Clearing the bell is not stopping anything. The notification center holds what happened until you’re ready and empties in one button, and that button is about the list of things you have seen. The instructions producing them carry on. Mark everything read and the watch fires again tomorrow, in exactly the same way. That’s the correct behavior, and it’s worth knowing before you clear the badge.
One question decides between the two levers. Would you want this in front of you at eleven o’clock this morning? If yes, keep the watch and change how it reaches you. If no, you’re in the next section.
What ending a watch actually stops
One click ends it. The watch is canceled, the bell goes quiet on that one, and no further notifications come from it. That’s the whole of it.
The cancellation is written down, timestamped, with a link back to the task, in the same record that holds everything else Point does. That matters more than it sounds for anyone who has ever had to account for their systems, because an instruction you retired in the fall can still be explained the following spring. The record of what a watch caught while it ran stays after you end it too. The activity log behind every action is where all of that lives and how to read it without turning it into a second inbox.
Then there are four things that carry on exactly as they were, and knowing them saves you clicking the wrong thing when it’s a different lever you want.
Ending stops future fires. Anything the watch already set in motion stays as it is. Reversing that is a separate action with a separate button, and undoing what Point did is the piece on how far that reaches and where it stops.
Work already prepared stays put. If the watch was carrying a next step, anything it prepared before you ended it is still sitting where it was, waiting on you, and you either release it or turn it down. That whole arrangement, and what prepared means, is when a watch lines up the next piece of work.
The dial stays where you left it. How much gets done for you is held per kind of action, sitting at review out of the box, and it’s set in one place and read in another. Ending a watch leaves every position where it is, and so did setting one. Setting how much your inbox does on its own is where those positions begin and end.
The mail still arrives, still sorted. The mail the watch was catching still comes in and still gets ranked with everything else. You have stopped being interrupted for it and you’re still receiving it, and how email triage works is the part still doing its job afterward.
One small thing that saves a click: a one-shot watch that has already fired has finished. Going to end it is tidying rather than stopping.
The twenty minutes that keeps the list honest
Standing instructions fail quietly. Nothing on your list will ever tell you that its reason expired, so the defense is a date rather than attention.
Once a quarter, open the list and read every row assigned to Point. On a healthy list this is fifteen or twenty minutes, and most rows survive it. For each one, three questions in order.
Does the reason still exist? Is the matter still open, is the client still a client, is the season still the season. This kills more rows than the other two together, and it kills them fastest, because you can answer it from memory.
Has it fired, and were the catches ones you wanted? The last three fires, read in the two minutes described above. Zero fires sends you to the search in the section before that.
Would you write this sentence today? Would you set it now, from scratch, knowing what the last three months were like. This is the question that clears out the watches set in a bad week for a worry that turned out not to be the real one.
Anything that fails one of the three gets ended, and the click is cheap precisely because the record survives it. The evidence stays and the future noise stops.
Two moments deserve a read outside the quarterly one. The first is when a client, a matter or a season ends, since that’s when several rows go stale at once and it’s the only time you’ll remember which. The second is when you change how much Point does on its own, because the same sentence can behave differently after a setting moves, and nothing prompts you about it. That consequence belongs to when a watch lines up the next piece of work, and it’s worth reading before you raise anything.
How many watches you should be running at all is a different question, with an answer of its own. The arithmetic of interruption, and why the casual ones make the careful ones worth less, is worked out next door. What this page adds is that the number stays honest only if somebody reads the list, and a list nobody reads grows in one direction.
When the thing pestering you is not a watch
Plenty of people arrive at a page like this having decided the software is telling them too much, and then end the wrong thing. Point’s list holds the instructions you gave Point. Four other sources of noise sit outside it.
The daily briefing. A digest on a schedule, with its own switch among the notification settings.
A thread that came back. Something you asked for and heard nothing about returns to the top of your feed on the day you named. It has its own controls, described in unanswered emails, and the threads that come back.
Mail other systems send you. The bank, the portal, the monitoring tool, the payroll provider. The sending happens at their end, so a Point watch you end leaves all of them running. When six systems mail you about the same event, the fix is a different one and it belongs to turning ten alerts into one.
Filters and rules you set years ago in your mail provider. These are the ones people forget entirely. Gmail keeps its filters in its own settings, in a section Google calls “Filters and Blocked Addresses,” and its help page for filters carries a section headed “Edit or delete filters” (checked September 6, 2026). Outlook keeps rules in its own settings too, and Microsoft notes that in classic Outlook for Windows one can be paused instead of removed: “To temporarily suspend a rule, clear the check box next to the rule. This enables you to keep the rule and turn it on later” (checked September 6, 2026). Both lists live with the provider, because Point holds what you asked Point for.
The diagnostic is quick. Read the notification you’re annoyed by. If it names a watch and links to the message that set it off, the lever is here and the last two sections cover it. If it’s an email from a machine, the lever is at the sending end, in whatever system wrote it. Everything Point itself can be asked to hold on your behalf is on the benefits page, and if you’d rather see the first one being set and then found again, day six: setting your first watch in plain words walks through it on the morning somebody did.
Common questions
How do I see all the watches I have running?
Open your task list. Every watch is a row in it, assigned to Point and marked as watching, sitting with your own tasks rather than in a separate automation screen. That includes watches on incoming mail and scheduled checks that go out to the web on a timer. If you’re trying to find one particular watch because it just interrupted you, go the other way and open the notification instead: it names which watch fired and links to the message that caused it.
How do I turn off an email watch?
Open the watch from your task list and end it in one click. The bell goes quiet on that one and no further notifications come from it. The cancellation is written into the activity log with a timestamp and a link back to the task, so the fact that you stopped it stays on the record.
Does stopping a watch delete what it already found?
No. The record of what it caught stays where it was, and the messages themselves were always just your mail. Ending a watch is about future fires only. If the watch had already prepared a piece of work that’s sitting on your list waiting for you, that stays too, and you release it or turn it down as you would have anyway.
I have a watch that has never fired. Is it broken?
Maybe, and one search settles it. Look in your own mail for the thing that should have set it off. If it hasn’t happened, the watch is fine and the silence is correct. If it happened and the watch stayed quiet, the sentence was probably written around words the other side never used, so widen it and set it again. If it happened and you found out some other way in good time, the watch was carrying no weight, which is a reason to end it rather than to fix it.
Can I pause a watch instead of ending it?
You have two levers: how it reaches you, and whether it runs. If the trouble is the interruption rather than the information, turn off the sound, the banner or the email copy and leave the watch working. If the trouble is the watch, end it, and write the sentence again in ten seconds on the day you want it back. There’s one position each way, and that’s deliberate: an instruction you can’t see running is worse than one you have to re-create.
Will ending a watch stop Point acting on my mail?
No, and the separation is worth being clear about. How far Point goes on your behalf is set per kind of action and sits at review out of the box, so work gets prepared and waits for you. A watch is an instruction about what you’re told. Ending one changes what reaches you and leaves the rest alone, in the same way that setting one left the dial where it was.
The short version
Every watch you have is a row in the list you already read, marked as watching and assigned to Point. Every row opens into the record that makes it answerable: your own sentence, whether it recurs, how often it has fired, and a link from each fire to the message that caused it. Read the last three catches to judge a noisy one. Run one search to judge a silent one, because zero fires means either nothing happened or nothing was ever going to. Change how it reaches you when the information is right and the interruption is wrong. End it in one click when the reason has gone, which leaves a timestamped line behind. Then put twenty minutes in the calendar once a quarter to read the whole list, because standing instructions fail quietly, and the only ones that go stale are the ones nobody looks at.