AI has quietly become part of my everyday life.
Some days it helps me build something I can already see in my head but haven't worked out how to make. Other days I use it for something I haven't touched in years, or barely understand, or know almost nothing about.
That second thing is where it surprised me most.
It doesn't make me an expert. It doesn't drop years of experience into my head. What it does is help me get further into something than I would have managed on my own.
That happened recently with AutoCAD.
I use AutoCAD regularly, but I'm no expert. And AutoLISP, the language you can use to build your own commands inside it, I had basically never touched.
I had a slow, repetitive job at work: take several drawing sheets, turn them into one PDF. Export one file, then another, then another, then open a separate tool to stitch them together. Every time I did it I thought there had to be a better way.
So I asked AI. Not because I knew the answer. That was the whole point.
I told it what I was trying to do, which version of AutoCAD I had, what I could work with, what wasn't working, and what I wanted at the end. Then we worked through it. The first version wasn't right, so I explained what happened and we changed it. Something else broke, so I went back, described it, tried again. Slowly, through that back and forth, we ended up with an LSP file I could load and use to do something I couldn't do before.
Now it saves me time every week.
But the interesting part wasn't the code. It was the conversation.
I've noticed the same thing with people trying AI for the first time. "I asked it to do something and it didn't work." "It made something completely different from what I wanted." A few attempts, some frustration, and they decide the tool can't do what they need.
Most of the time I don't think that's the problem. The problem is that the idea is still sitting inside our own head. We know what we mean because we can see the whole picture. AI can't see it unless we hand over the pieces.
It's not so different from talking to a person. If I say "you should have seen what happened with John last weekend," a close friend laughs, because they know John and already have half of it pictured. A stranger has nothing. Who's John? Where were you? Why was it funny?
AI is the stranger. I have the picture. It doesn't. So I give it the picture.
And this is the part I'd have found useful to hear earlier: you don't need to be a prompt engineer. If I don't know how to explain something, I tell it that too.
The line I use most is something like:
"I need to build something that does this. I'm not sure what else you need to know, so ask me questions before you start."
That one sentence changes the whole conversation. Instead of writing one perfect instruction and hoping for magic, I let it help me define the problem, not just solve it. When it gets something wrong, I don't decide it's useless. I tell it what's wrong, what I expected, and whether it's missing anything.
The other thing I've noticed is that it gets easier over time. Some of that is the tools remembering context. But most of it is me. I've got better at describing the problem, at knowing when more detail is needed, at saying "no, that's not what I meant."
It feels a bit like getting to know someone. At the start you explain everything. Later it flows, because both the context and the way you talk to each other have caught up. Not because it became my best friend. Because I learned how to work with it.
And there's a lesson in that which has nothing to do with AI. Explaining what you need, listening, correcting a misunderstanding, giving context, that matters everywhere. At work, in a team, with customers, with the people close to you. Good communication mattered long before any of this arrived. AI just makes it very obvious when I haven't explained myself properly.
So AI hasn't replaced what I know. It's done nearly the opposite. It's helped me reach into things I'd have left alone because I thought I didn't know enough. Coding, automation, AutoCAD, and plenty of smaller things in between.
I still have to think. I still have to test. I still get things wrong, and so does it.
But my first thought used to be, "I can't do that."
Now it's turning into, "Okay. How could I learn enough to try?"
That's a very different place to start from.
No noise. Just open talk. No shortcuts.