You Don't Have to Do It Yourself Anymore. You Just Don't Know It Yet.

Setting up the same Excel sheet from scratch every month, entering the days, copying the hours over. Sound familiar? You don't have to anymore. Fire up an AI, describe what you need, and you've got a finished tool. You can hand it straight to someone, and it just works.

If you work with AI every day, that takes a few minutes. For everyone else, it probably doesn't even come up as an option.

You think everyone knows#

When you work with AI every day, at some point it all feels completely normal. You're constantly handing things off to agents, building tools for all kinds of stuff, and everywhere you look it's new models and new features. So you kind of automatically assume everyone roughly knows what's possible right now.

They don't. You're sitting in a pretty thick bubble and don't even notice. At the latest when you talk to people who don't work with this stuff every day, you realize how far apart the two worlds are. Gut feeling: 95% of people have no idea what AI can already do today.

Not really surprising, actually. The models are so powerful by now that even you barely keep track of everything you could use them for.

What already works today#

One example: I wanted to mark a specific spot on an image, briefly described it to the model, and it marked it. In the right place. Looks like a gimmick, but for that the model has to understand what's in the image and where exactly the thing I mean is. And it goes further: throw in a photo of a part, a car, or a wall, ask "Do you see any damage?", and the model marks scratches, cracks, dents, or rust. That used to be a project of its own with a specially trained model. Today it's one sentence in a chat.

And it goes like that all through everyday life:

• Writing: Need a letter to cancel a contract? You can sit down for five minutes and think about how to word it. Or you just say: "Write me a cancellation letter for XY."
• Documents: Build a PDF form without learning some program first.
• Organizing: Clean up your inbox, check your calendar, dig things up.
• Free time: Have a little tour planned on the fly while you're on vacation.
• Your own software: Mini tools like a time tracker, your own blog website, or my little workspace where I manage several projects and work with coding agents. All things that used to take days or weeks.

The old mindset#

What a lot of people still have in their heads: I have to do this myself. I have to know this myself. I have to learn it first.

You really don't anymore. You should know roughly what you're doing, but not every detail. Your role shifts: away from doing it yourself, toward asking, checking, and deciding.

And the checking really is part of it. The models get things wrong sometimes, but the direction is usually right. So don't take things over blindly. Look at it, dig deeper, and if in doubt, ask again. That's still way faster than working everything out yourself.

Sure, at some point it stops. When it comes to really big, complex systems with lots of interfaces and dependencies that need to stay maintainable and extensible for years, you do need someone who knows what they're doing. But for a small everyday tool? It's totally fine.

So what now?#

I'm not saying everyone has to do everything with AI now. But it's worth stopping for a second: Where am I still slogging through stuff I could hand off in a few minutes? And what am I only doing this way because I've always done it like that?

Next time you catch yourself thinking "Ugh, I have to do this by hand again": just quickly ask an AI if there's another way. Most of the time, there is.

And where could all of this lead one day? I've got some ideas. More on that in the next post.

---

PS: Yes, I worked on this text with AI again. Like last time, I just started talking and then had Claude interview me. You get questions you wouldn't have come up with yourself. The thoughts are still mine. I just skipped the typing.