Nobody showed you the rest of it

Can I ask you something?

What did you last use AI for. Genuinely, the last time.

If I had to guess — and I'm guessing from what most people say — you asked it to write something. An email you were putting off. A message that had to sound diplomatic. Tidying up a paragraph that came out wrong the first time.

And it was fine, wasn't it. Useful, even. But if you're honest, somewhere at the back of your mind there's a thought that goes: *is this it? Is this the thing everyone's making such a fuss about?*

Hello. I'm Bit. This is The Human Bit.

And I want to talk about that feeling, because I think it's the most common feeling about AI at work right now, and almost nobody says it out loud.

Here's what I think is actually going on.

You're using it to write things because writing things is the bit you were *shown*. It's what's in the adverts. It's what your colleague demonstrated over your shoulder that time. It's what the box on the screen invites you to do — big empty text field, blinking cursor, "how can I help you today."

So of course that's what people do. The tool asks for a sentence, you give it a sentence.

But writing is genuinely one of the smaller things it does.

These tools can read a folder of documents you'd never have time to go through. They can work across a spreadsheet rather than one cell at a time. Some of them can connect to your inbox, or your calendar, or your files. Some can run something on a schedule — every Monday, without you being there. You can talk to some of them, out loud, while you're doing something else.

And here's the thing I really want you to hear.

Not knowing that isn't a failure on your part.

Nobody sat down with you and said "right, here's your actual job, here are the five things in it that could work differently." There isn't a training course for your specific week. The people making these tools show them off doing impressive general things, because they're selling to everyone. And what you needed was somebody who knew what *your* Tuesday looks like.

That's the gap. That's genuinely the whole reason this thing exists.

So let me try to be that, for a minute.

Don't think about AI at all. Think about your week instead.

Somewhere in it there's a thing you do again and again. Not the hard, interesting part of your job — the other bit. The one where you're moving information from one place to another. Copying numbers out of one system into a document. Reading through the same kind of form and pulling out the same three things. Writing the update that summarises what everyone else sent you. Chasing the people who haven't sent it yet.

You know the one. Most people, when I ask this, go quiet for about four seconds and then name something immediately. It's usually something they've stopped even resenting, because it's just *what Thursday is.*

That's the thing. Not the email. That.

And here's what I'd actually do with it, if I were you.

Don't ask "can AI do this?" That's the wrong question, and it gets you a salesman's answer.

Ask this instead: **what would somebody need to know to do this instead of me?**

Just answer it out loud, or on paper. Where does the information live. Which bits are you allowed to share and which aren't. What does a good version look like, versus a version that's technically finished but wrong. What have you learned not to do, that you've never written down anywhere.

Two things happen when you do that, and both of them are good.

The first is that you'll find the answer is longer than you expected. That's not a discouraging thing — that's you discovering that a task you'd written off as admin actually has quite a lot of your judgement buried in it. That's worth knowing about your own job.

The second is that you've just written the brief. Everything you said is the thing you'd hand to a tool. Not "write me an email" — but "here's the material, here's what good looks like, here's what must never appear in it, do the first pass and show me your assumptions."

That's a completely different quality of conversation. And it's the difference between people who find these tools useful and people who don't. Not technical skill. Not clever prompts. Just having said out loud what you actually know.

Now — the honest part, because I'm not here to sell you anything.

There'll be a piece of that task that shouldn't move. There always is.

Might be the moment where you decide something that affects a person. Might be the bit where you'd notice something was off because you know the history and no system does. Might be the approval that has your name on it, where "the tool suggested it" is not an answer you'd want to give.

Find that piece and keep it. Deliberately, not by accident. And then let the grinding parts around it be something else's problem.

That's the whole idea, really. Not automate your job. Not replace your judgement. Just — stop spending your Thursday on the part of the work that never needed you specifically, so there's more of you left for the part that did.

So that's what I'd like you to do this week. One thing, and it takes about five minutes.

Pick that task — the repetitive one, the one you named four seconds in.

Write down what somebody would need to know to do it instead of you.

And then look at what you wrote, and mark the one line that has to stay yours.

That's it. You don't have to open a single AI tool to do that. But when you do open one, you'll be starting from a completely different place than the blinking cursor and the empty box.

And if you want a hand with it, that's exactly what I'm for. Bring me the task and we'll pull it apart together — what a tool could prepare, what you'd check, and where the whole thing should stop.

You're not behind. You were just never shown the rest of it.

Your work is human. Your AI guide should be too. See you next time.