You have decided, so this page will not spend a paragraph arguing. What is in front of you now is an afternoon, and it is a much smaller afternoon than the decision was. You pick Gmail or Microsoft 365, you sign in to your provider the way you already sign in to it, you read one screen and agree to it, and then you wait a while for something to read back through eleven years of mail. There is no export, no import, no forwarding to configure, no new address, and nothing to tell your clients.
So the real question of the first day is not how. It is what to do with yourself for the hour afterwards, and what to leave strictly alone until you have watched the thing work. Both of those have better answers than most people give them, and getting them wrong is why some perfectly good first weeks end in a shrug.
Here is the day, in the order it happens.
What the couple of clicks actually are
You start inside Point and you finish inside Point, but the middle of it happens somewhere else, which is the part worth understanding before you get there.
Choosing Gmail or Microsoft 365 hands you over to Google’s or Microsoft’s own sign-in page. That is their page, not a copy of it, and your password is typed into the company that already has it. Point never receives it and has no box to put it in, which is a smaller reassurance than it sounds and is taken apart properly in why Point never needs your email password.
What comes next is the screen that matters: a consent page, listing what the application is asking to reach. Read it once, slowly, because that list is the agreement. Everything Point is able to do for the next two weeks, and everything it is not, is described there in your provider’s words rather than in anybody’s marketing. If you want it set out in advance so the screen holds no surprises, what Point can see, and what it cannot is the same list with the reasoning attached.
Then you agree, and you are back in Point, and the doing is over. Point’s own sign-in, the one you will use tomorrow morning, is a link sent to your address rather than another password to keep.
One thing to settle before the afternoon rather than during it. If your Workspace or Microsoft 365 is administered by somebody other than you, they may have a say in whether a new application can connect at all, and that is a short email to send on Monday instead of a wall to hit on Thursday. If your day involves a shared address like accounts@ or a mailbox somebody else manages on your behalf, those have their own answers and they are not the same answer: how to switch to an AI email client works through both, along with your signature, your other sending addresses and your phone.
The part with the waiting in it
Point now reads what is already there. Not the last month and not a sample: the mail sitting in the account, going back as far as it goes back, because deciding what matters is exactly the same activity as reading it.
This is the only part of today with any real wait in it, and how long it takes depends on how much is in there. Nothing is being copied out to make it happen and nothing is being moved: your provider still holds the mailbox, still delivers to it, and still charges you for it. If a message lands while the read-back is running, it is triaged on arrival rather than at the next refresh, which is the ordinary behaviour from here on and worth seeing described properly in triaged the moment it lands.
The advice for the wait is to leave. Go and do an hour of real work in the inbox you have always used, which is still sitting there working exactly as it did this morning.
That is not restlessness management. The feed assembling itself while you watch is the least representative view of Point you will ever be shown. It is a mailbox at rest, ranked from history, with nothing live in it, nothing learned about you yet, and you leaning over it waiting to be impressed or disappointed. Judgments formed in that hour tend to be strong and tend to be wrong in both directions. People decide the software is uncanny because it surfaced something they had genuinely lost, and people decide it is useless because the fourth item down should have been the second. Neither of those is a measurement. Both of them are a first impression of a machine reading old mail.
So let it finish while you are elsewhere.
Go and open your old inbox
When it is done, do the thing nobody thinks to do on the first day, and open Gmail or Outlook directly.
Connect Gmail and Point creates three labels there: Now, Later and Other. Every message it has triaged carries the one it earned, so a single click filters your real mailbox down to what needs you, inside Gmail itself, with Point closed. Connect Outlook and the same three arrive as categories, doing the same job in the same way.
Spend two minutes on this, because it settles something that the feed cannot. What you were half afraid of when you connected was becoming dependent on a window: that the sorting would live inside somebody’s application, and that a bad month or a bad decision at that company would take your working order away with it. It does not live there. The judgment is written back into the mailbox you have kept for a decade, in the mailbox’s own vocabulary, readable from the app you were using yesterday and from your phone before you have installed anything.
The writing back goes both ways from here. Put a thread away in Point and the real mailbox agrees a second later: out of the inbox, archived rather than deleted, sitting wherever the rest of your archived mail sits. That is the mechanism behind a great deal of what follows, and it is the reason the two apps do not drift apart while you are still running both. It is also the honest version of what happens if you ever stop: the labels stay, the archived mail stays archived, and the switching guide is straight about what does not stay, which is the layer on top.
The one screen to find before the feed
Before you let yourself read the sorted feed, find the autonomy dial.
The order matters more than the setting does. Find it now, while you have no stake in the software doing well, rather than in a fortnight when it has done three impressive things and you would quite like to give it more rope. Ten minutes from now the feed will have led with a message you had forgotten about, and you will be a slightly different reader of a settings page than you are at this moment.
What you will find is not one switch. It is a level held separately for each kind of work: filing, drafting, scheduling, chasing. Each has three positions, from Point staying out of it, to Point preparing something for you, to Point doing that work itself, and each of them arrives in the middle one, the one called review: Point prepares, you approve. So on the afternoon you connect, nothing reaches a client and nothing leaves the building without passing under your eyes first, not because a promise says so but because that is where the settings sit until you move them. Setting how much your inbox does on its own is what each position does and where its reach ends, if you want the whole of it now.
Then leave every one of them exactly where it is.
The reason is not caution, and it is worth being precise about, because caution is the reason people give and it produces a worse fortnight. The reason is measurement. You are about to spend two weeks forming a view of how good this thing is, and a view is only worth having if you can tell what produced it. Move one setting today and one tomorrow and something odd on Friday has three possible parents. Move nothing this week and every surprise has a single cause you can name. There will be a right moment to hand something over, and it is not eight minutes after you met the software.
What to leave alone until tomorrow
The same logic runs through everything else you might be tempted to tidy this afternoon, and there is more of it than you would think.
Leave your filters and your rules. Every one of them is still running, because the server that runs them has not changed, and the tidying instinct on day one is strong. Resist it for a fortnight. You are about to get evidence about which of them the ranking has quietly made redundant, and pruning with evidence is a different exercise from pruning on a hunch. There is one specific trap here, and it is worth knowing rather than discovering: a filter that archives mail on arrival will keep hiding it, so if a particular client’s messages are not turning up where you expect on Wednesday, an old rule is the first suspect and the software is the second.
Leave the old app on your phone, and do not install anything new today. There is a version of Point that lives in your pocket, and a swipe there files a thread in the real mailbox as well, but the phone is a second variable and today already has one. The only thing worth doing about the phone this week is making sure two mail apps are not both buzzing at you, which is an irritation people reliably blame on the new software.
Leave the marking of who matters. Point will arrive with your contacts, merging the duplicates that have accumulated over the years, and there is a way to name the handful of people whose mail should never end up in the middle of a list. It is worth doing and it is coming later in this fortnight, deliberately, because there is something to learn first from watching what the ranking gets right about your clients before you have told it anything. Making sure the people who matter are never buried is the mechanism, for when the moment comes.
The one thing that is genuinely better done now than later is a second business, if you run one. Connect both accounts this afternoon rather than adding one in week three, because the difference between the two matters to how everything is sorted from the first message onward, and keeping two businesses apart is what that separation actually amounts to.
The afternoon proves nothing, and that is fine
Read the feed now, if you like. You have earned it, and it is the thing you connected for: the messages that need you at the top, each one carrying a plain line telling you what is in it before you open anything. Everything Point does is the full inventory behind that view, and most of it starts working today whether or not you go looking for it.
But hold the verdict, because the afternoon is not evidence and it was never going to be.
Everything you have seen was Point reading a mailbox at rest, on a day you chose, with you sitting over it. Nothing arrived that mattered. Nothing had to be decided under pressure. You already knew what was in there, which means the only thing you could really assess was whether the software agreed with a ranking you were carrying in your head before you sat down. That is a test of agreement, not of usefulness, and it is why first afternoons produce such confident opinions and such unreliable ones.
So close the laptop without tidying anything. Do not sort the feed into an order you like better. Do not correct the four items you would have ranked differently. Leave it as it is, with all its judgments intact and none of your fingerprints on it, because tomorrow is the first honest measurement you get.
Tomorrow, mail arrives while you are asleep. Nobody watches it land. The ranking is made by something that has never met you at half past six in the morning, on messages you have not seen and cannot predict, and you come to the result cold, with a coffee, on a working Wednesday with clients in it.
And here is the thing that actually happens on that morning, which nobody warns you about and everybody does. You will read the feed, and then you will open Gmail anyway, and scroll the real list to the bottom to check that nothing was hidden from you. Not because you saw an error. Because you have not yet been given a reason not to, and forty minutes of sorting a day is a habit with roots deeper than one afternoon.
That second look is the right instinct on the second day, and it is the whole subject of it.