I am not a graphic designer. I have said so on the show more than once, and I meant it both times. For years, my process was to open Canva, scroll until something looked less bad than everything else, and drop my text into it.

So when image generation got good, it looked like the answer. Describe the thing, receive the thing, ninety seconds, done. The trouble is that the thing arrives finished. You want the headline two points bigger, and now you are negotiating with a model that moves something else every time you ask.

Then the platforms started attaching labels. That little tag naming your post as AI content shows up on the one piece of the week where you wanted someone thinking about your product instead of your production method. You saved an hour and bought a conversation you did not want to have.

Peg Fitzpatrick found the way around it, and the way around it is not a trick. She simply stopped asking for an image.

Peg runs marketing for a company that sells industrial textile care products, which is to say she is in-house, she is accountable for the brand, and she does not get to experiment on it. Her rule is absolute. No AI-generated imagery for a brand she is responsible for.

What she built instead is a Claude skill that produces a campaign and then hands the design work to Canva as a Canva file. Claude writes the captions, the story copy, the Pinterest fields, and the LinkedIn variations.

Source link