AI is everywhere in creative work right now. It writes, it designs, it suggests, and it does all of it quickly. We’re curious about it the way we’re curious about any new tool, and we’re paying attention to where it goes.

We’re also a little wary. Not of the tools, but of what they make easy.
If you run a business or an organization, you’re probably using these tools too, or being handed work made with them. So this is less about how we work and more about what to watch for.

Twenty directions, zero opinions

Ask an AI for twenty directions and you’ll have them before your coffee cools. Most will be fine. A few will be interesting. None of them will have an opinion about which one is right.

That isn’t a flaw. It just isn’t the tool’s job. Choosing means knowing things that aren’t in the prompt: that the people reading this will mostly see it on a phone, in between other things. That the word everyone in the meeting keeps using means nothing outside the meeting. That the organization tried something similar a few years ago and has its own reasons it didn’t stick.

Most of that never makes it into a document. It comes out in conversation, often sideways, in a comment someone makes on the way off a call. A tool can’t overhear that. Someone has to be listening.

Design is a pile of small decisions

From the outside, design can look like one big decision: the logo, the homepage, the cover. In practice it’s hundreds of small ones. What goes first. What gets cut. Which number in a forty-page report is the one people need to remember. How much a reader can take in during the few seconds they’ll give it.

Each of those decisions is minor. Together they’re the difference between something that works and something that only looks finished. You can make a website that’s impressive and hard to use, or a beautiful report nobody reaches the end of. The software didn’t cause either one. A decision did, or more often, a decision nobody stopped to make.

The risk is skipping the stop

That’s the part we watch most closely. AI doesn’t make bad work on its own. What it does is make it very easy to skip the moment where you stop and ask whether this is right.

When a draft appears in seconds, it feels done. It has paragraphs and structure, and it sounds sure of itself. When a layout arrives already arranged, the instinct is to polish it rather than question it. Speed builds momentum, and momentum is hard to argue with.

It’s a pull we feel too. The first version that shows up isn’t always wrong, but it’s rarely the best one, and it’s harder to tell the difference when you didn’t have to work to get there. Sometimes the effort of getting to an answer is how you find out it was the wrong question.

The answer isn’t avoiding the tools. It’s protecting the slow parts: asking one more question about the audience, rereading the brief and noticing it describes a different problem than the one the organization is actually facing, putting three rough versions in front of a real person instead of perfecting one in isolation. Those parts don’t get faster, and they shouldn’t.

Sometimes the right tool is something we didn’t have last year. Sometimes it’s a pencil. Part of the job is knowing which, and being honest when the simpler approach is the better one.

Before you publish something AI helped make

Write the point down first. Before you open any tool, write one sentence: who this is for, and what they should know or do after seeing it. If you can’t write that sentence, the tool can’t either. It will just sound like it can.

Read it out loud. AI drafts tend to sound confident and say very little. Listen for sentences that could sit on anyone’s website. Either make them specific to you or cut them.

Check every fact. Names, numbers, dates, prices, claims about what you offer. AI fills gaps with things that sound plausible, and it doesn’t flag which parts it made up.

Look for what’s missing. The tool gives you the expected version. What it can’t give you is the thing only you know: why you started, the detail your customers always mention, the way your community actually talks. If none of that is in there, it could belong to anyone.

Show it to someone outside the project. Don’t ask if they like it. Ask what it’s about. If their answer doesn’t match your one sentence from the start, it isn’t finished yet, however finished it looks.

What doesn’t change

AI can be a powerful tool, but it's not what makes great creative work. What matters is the creativity, experience, and judgment behind it. Technology helps us explore new possibilities, but thoughtful decisions and a clear understanding of the problem are what make the difference.

That has always been the work. Understanding what an organization is trying to do. Understanding the people it wants to reach. Finding the idea that connects the two, and making it well enough that it actually lands. AI can help with a lot of that. It can’t tell you whether you got it right.

Faster is nice. What we care about is whether the thing works for the people it’s for, and no tool has made that question go away.