Nobody will care that you used AI, but they’ll care if your writing sucks

Some weeks I lose my voice. Apparently, talking to yourself all day will do that.

I write by talking now. On walks, over the dishwasher, halfway through a pile of laundry, and yes, during the occasional shower idea. I felt ridiculous for about a week. Then it became the most natural thing I do all day.

It makes sense. My best thinking never happened at a keyboard. It happened when my hands were busy and my brain wandered off.

By the time I sit down, the thinking is done and the words just have to catch up. Give AI some context about what I’m writing and how I sound, and the draft comes back sounding a lot like me.

That’s what I expected. What I didn’t expect was how much it changed my idea of what writing is.

The words were getting easier. The thinking wasn’t.

And I think the same thing is about to happen to content design.

Everybody can draft

Think about product managers. They can write a headline…of course they can. Just like engineers can fix an error message. Designers can generate ten options before lunch. AI can, too, only faster and arguably better. It can summarize research, apply voice and tone, check terminology, and it’s on track to keep eating bigger pieces of the workflow.

So saying “I’m the person who writes the words” gets weaker every quarter.

“I’m the person who understands what we’re trying to say, why, to whom, in what context, and what happens if we get it wrong” gets stronger.

I’ve spent a few years working out how to write with AI, with my team and on my own. I landed on five rules I originally wanted to share as my guidelines for AI writing. But rereading them now, they aren’t really rules about writing. They’re more like a preview of where the job is going.

1. Different work deserves different levels of AI autonomy.

An email, a note to the team, a LinkedIn post, a script for a deck. I tend to use the same tool, Claude, to do four very different jobs in very different ways.

Some work can run on its own. Some needs review. Some needs a human all the way through.

Email and meeting notes can usually rip solo. That’s compression. But you want to come over the top afterward saying the one thing that matters most, because AI summarizes everything and prioritizes nothing.

Other pieces need you in the room the whole time. So “should AI write this?” is the wrong question. Ask what the piece is for, how much you can hand off, and where you’re needed.

A friend told me about a banner. One banner, pointing customers to a landing page that fixed their problem. Low risk, easy to change if it flopped. Five layers of VP-level leadership reviewed it. AI could’ve written that banner in seconds and checked it against every standard the team had. It still would’ve traveled through five VPs, because somewhere along the way visibility turned into review and review turned into approval, no matter the risk.

That’s the part AI won’t fix. When execution gets cheap, deciding who owns a decision and what deserves review becomes the expensive part. Content designers make those calls every week. We just haven’t given the work a name.

2. Everybody can draft now.

Your engineering partner can spin up a headline if you forgot one. Your product partners are surely writing feature content if you’re not fast enough. Soon nobody will care that you used AI or who used AI. They’ll care if your writing sucks.

And soon organizations won’t care that a content designer can produce excellent copy either. They’ll expect it. Excellent copy becomes the expectation.

When everyone can draft, you can’t defend your value by owning the words. The bar moves to framing the problem, understanding the customer, seeing the whole system, and knowing what good looks like before anyone writes a line.

Solving the right problem. That’s where our value moved. And it sounds exactly like content design.

3. Laziness reads as laziness.

When the AI tells are obvious and could’ve been fixed in seconds, people don’t care that you used AI. They care that you didn’t bother.

If you couldn’t be bothered with the words, why would anyone trust the thinking underneath? I wouldn’t.

AI makes judgment visible. Every draft now comes with a decision attached: did you read it, did you fix it, did you push back? Knowing what to delete, what to question, what to push back on, and what to reject outright has always been an important part of our craft. And it’s finally the part people can see. 

4. Longer is usually worse.

As the old saying goes, I would have written a shorter letter, but I didn’t have the time.

Don’t let your writing run long because you didn’t take the time to make it short. Cut a quarter at a time and watch your best lines finally show up.

This rule gets bigger with AI. Another paragraph costs nothing. Another variant. Another nudge. Another personalized message nobody asked for. AI makes it cheap to generate more of everything, and, good grief, look at your inbox.

Good content design has never been about filling every available space. It’s making the most of the space you have.

Sometimes the best content decision is silence. Sometimes someone has to say “we shouldn’t say anything here” or “Why would we do this?” That someone is us.

5. Writing is thinking, and you can’t hand off the thinking.

This is why talking with Wispr works. By the time I open my mouth on a walk, I’m doing the hard part: deciding what I actually mean.

AI is strong in the middle. But the ends are yours.

Stretch that across content design and it holds true.

The beginning: What problem are we solving? What does the customer need? What should happen next?

The middle: Research, synthesis, drafts, variations, analysis, QA. More and more of this is AI. And that’s okay.

The end: Is this right? Does it solve the problem? Should it ship? What did we learn?

Human judgment sits heavy at both ends and that’s where solid craft shows up. Human effort in the middle is shrinking. The amount of work happening there isn’t. It’s getting bigger and less visible, which is exactly why the ends count more than they ever have.

From writing words to designing behavior

Today the work looks like this: a screen shows up, a content designer writes the words.

Increasingly it looks like this: a customer says what they want, an AI tool interprets it, reasons about it, does something, and explains itself along the way.

The content designer doesn’t write every response. They can’t. There are millions. What they do is define what good behavior looks like.

What should it understand? What can it do? What can’t it do? When should it explain, and when should it just act? When should it ask? When should it hand off to a person? What should it sound like? And how will we know it worked?

Content designers don’t write a script for AI agents. We teach them how to talk. And talking, for an AI tool, includes knowing when to act and when to stop.

That’s a much bigger future than “AI-assisted UX writing.” It’s also harder, because you can’t hide behind a beautiful sentence anymore. But, then again, the sentence was never the point.

Back to the walk

Maybe that’s why talking to AI while I walk has changed my writing so much.

I’m not outsourcing the thinking. I’m separating the thinking from the production.

Content design is about to go through the same separation. AI will produce more and more of the experience. Our job is deciding what it should do and say in the first place.

That’s why, instead of sitting down looking at a Figma screen, I’ll be out walking, talking to myself like a lunatic, hoping I don’t lose my voice again.