Most people paste everything into AI by hand. They copy an email in, ask for a reply, copy the reply back out. It works, but it is slow, and the AI only ever sees the one thing you pasted.
There is a better way, and it is not technical. You let the AI read from the apps you already use. Then it can answer with your real inbox, your real calendar, and your real files, instead of guessing from whatever you happened to paste.
The thing that makes that possible is called a connector. Here is what it actually is, the few to turn on first, and the privacy check to run before you connect anything.
What a connector actually is
A connector is permission. That is the whole idea.
When you add a connector, you are telling the AI: you are allowed to look at this app of mine. Once you do, it can read from that app to answer you. Ask “what is on my calendar Thursday” and it can actually check, instead of asking you to tell it.
You are always in control of this. You choose which apps it can see. You can turn a connector off as easily as you turned it on. And a connector is not the AI wandering through your accounts on its own. It reads when you ask it to, for the thing you asked about.
One important distinction, because it matters for safety. Some connectors are read-only: the AI can look but cannot change anything. Others let the AI act: send the email, add the calendar event, edit the file. Reading is the gentle version. Acting is the powerful version. You want to start with reading.
The Connected Setup: start with three
You do not connect everything. You connect three things, in this order, and you stop there until it feels natural. I call it the Connected Setup, and the whole point is that three is enough to change how AI helps you.
Here is the order I give people, and why.
| Order | Connect this | Why it is first | The win it unlocks |
|---|---|---|---|
| 1 | Calendar | Lowest risk, mostly read-only, instantly useful | ”What does my week look like, and where are the gaps?“ |
| 2 | Highest daily payoff once you trust it | ”Summarize what I missed and draft the three replies I owe.” | |
| 3 | Docs and files | Turns your own writing into something AI can use | ”Pull the numbers from last month’s report into a summary.” |
Calendar first, because it is almost always read-only and the payoff is immediate. Email second, because it is where most of your day lives, so the help is huge, but the stakes are higher. Files third, because once the AI can read your own documents, it stops giving generic answers and starts giving yours.
That is the Connected Setup. Three apps, in order, read before act. Everything past that is a bonus you add later, one at a time.
Once those three feel natural, a few other connectors are worth knowing about, no deep dive needed: Slack (so it can read and summarize team threads), Notion (so it can pull from your team’s docs and wiki), Google Sheets (so it can work from a live tracker instead of a pasted screenshot), and HubSpot (so it can pull from your CRM if you run marketing or sales). Same rule applies to all of them: read-only first, one at a time.
What this looks like when it works
The AI I run every day is wired into my real tools: calendar, Gmail, and Google Drive for daily life, Slack and Notion for work, Canva and Figma for design, plus GitHub, my CRM, and my social accounts. I added them one at a time, read-first. Two of my AI systems are live in production, connected to real data the same way. That wiring is the whole difference. Disconnected, AI gives you clever answers about nothing. Connected, it works from your actual week, your actual inbox, your actual files. Here is what changes once it is there, in plain terms.
Before, at work: you open five threads to figure out what you missed, then write each reply from scratch. After: “Catch me up on what came in overnight and draft replies to anything that needs one, in my usual tone.” You edit and send. Fifteen minutes becomes three.
Before, at home: you scroll your calendar and your email trying to remember what the weekend actually holds. After: “Look at my calendar and my inbox and tell me what I have committed to this weekend, and what I still need to plan.” Now it can see both, so it can actually answer.
That is the shift. The AI stops being a clever stranger you brief every time, and starts being something that already knows what is on your plate, because you let it look.
The one prompt to run first
Once your calendar and email are connected, this is the prompt I would run before anything else. It turns the connection into a daily brief you actually use. Paste it in and adjust the wording to fit your day.
Look at my calendar and my email for today.
1. Summarize my calendar: every meeting, the time, and what I need
to walk in ready for.
2. Flag the 3 emails that most need a reply from me today, and say
why each one matters.
3. Tell me the one thing I should protect time for that nothing
on my calendar is holding yet.
Keep it short. Bullets, not paragraphs.Run it with your coffee. You get a one-screen picture of the day instead of clicking through five apps to build it yourself.
The privacy check before you connect anything
Run this every time, before you turn on a connector. It takes a minute and it is the difference between safe and sorry.
Ask three questions.
What can it see? When you add a connector, the tool tells you what it is asking for. Read that screen. Is it your whole inbox or one label? Read-only, or read-and-send? If the permission is broader than the job, that is worth noticing.
Read or act? Start read-only wherever you can. Let the AI look before you let it send or edit. You can always add the ability to act later, once you have seen how it behaves and you trust it.
Is this a work account? If yes, stop and check your company’s policy before you connect. This is not optional and it is not yours to decide alone. Some companies have an approved way to do this. Some say no. Find out which before you click connect.
One more note, and it is the honest one. These features change constantly. The menu names, the exact permissions, which tools connect to what: all of it moves. So do not memorize the buttons. Learn the concept, and check the current settings each time you set one up. The concept is evergreen. The screenshots are not.
The common mistakes
I have watched people, and myself, hit every one of these. Here they are so you can skip them.
- Connecting everything at once. It feels efficient. It is not. You lose track of what has access to what, and one messy setup sours you on the whole thing. Connect one, use it for a week, then add the next.
- Not checking what data it can access. People click “allow” without reading the screen that says what they are allowing. That screen is the whole safety story. Read it every time, even when you are in a hurry.
- Ignoring your employer’s policy. Connecting a work account to an outside tool can violate a policy you did not know existed. The fix is free and takes one message: ask before you connect. See how to set an AI policy for a team if you are the one people are asking.
- Connecting sensitive accounts too early. Your bank, your health portal, anything with real consequences. Those are not where you learn how connectors behave. Learn on your calendar. Keep the high-stakes accounts out of it until you are sure, and often keep them out entirely.
- Giving it permission to act before you trust it to read. Read-only first, always. A tool that can only look cannot send the wrong email. Earn the “act” permission slowly.
Do this today
Open your AI tool of choice. Find the connectors or integrations menu. Connect one thing, your calendar, and nothing else. Then ask it: “What does my week look like, and where are my open blocks?”
That one connection is the whole idea in miniature. The AI can see something real about your life now, so its answer is about you instead of about no one. Once that clicks, add email, then files, in that order.
If you are still getting comfortable with the basics, this pairs well with your first week with AI. And once your AI can see your apps, the next thing worth doing is teaching it who you are, which is exactly what giving AI a memory covers.
Want my exact connector setup, the ones I turn on first, and the privacy settings I always check before I connect anything? Grab it below and I will send it over.