I run a small T-shirt shop called Cutlery Crisis. AI rebuilt the website. AI sorted the SEO. AI helps with the marketing.
And there's one job it's not allowed anywhere near: the drawings.
Every design on every tee, hoodie, mug and sticker is drawn by hand. Mine. That isn't nostalgia, and it isn't a protest. It's the most commercially sensible decision in the whole business, and it's the clearest example I've got of where a designer's value lives now.
A shop that needed a face
Cutlery Crisis makes soft, funny clothing for the chronic illness community. The name comes from spoon theory: you start each day with a handful of spoons, everything costs one, and some people start with far fewer than others. Nobody can tell by looking.
The brand runs on two characters. Duckling is the face you put on: composed, coping, thriving, allegedly. Fuzzy is what's going on underneath: exhausted, hangry and out of spoons. The tagline sits between them: Looks fine. Isn't.
I started it after my partner became ill with several chronic illnesses, including ME/CFS and POTS. We wanted to raise awareness of chronic illness, and to put our own sense of humour into the designs: something to make people smile if they get it.
What the machine does
Almost everything that isn't a drawing.
The shop. This is the part that's changed the most. I started Cutlery Crisis over 18 months ago, before AI was much use for this kind of thing, so I built the first site myself. It was hard. I had to learn a lot, and every week I found something else I didn't know I needed.
The brand has been through two complete redesigns since, as I tried out different styles. The last full rebrand was built by an AI coding agent from a brief I wrote, the same way I built the landing page for my book. It took a couple of weeks to get the whole site up to speed. It was ready well before my new designs were, which isn't the worst problem to have.
The SEO. Product titles, meta descriptions, alt text for every mockup, the plumbing that helps someone searching for "chronic illness clothing" actually find the shop. This is exactly the kind of patient, repetitive, rules-based work AI is brilliant at, and that I was never going to do well at 11pm.
The marketing. Drafting product copy, planning posts, writing the newsletter. AI does the first pass; I keep the voice. The shop's tone is deadpan and a bit tired (the newsletter sign-up promises updates "when we have the energy"), and a model left to itself drifts straight back to cheerful-brand-speak.
The fulfilment. It's print on demand: Shopify connected to Printful, so there's no stock and no warehouse. When someone orders, Printful prints the design and posts it. No AI needed here, just good plumbing.
Put together, that's a whole shop's worth of operations that a single person could not otherwise run around a day job.
Why it doesn't get the pencil
I could generate a hundred cartoon ducks before the kettle boils. I've tried. They're fine. They're also exactly what everyone else's shop is selling.
Here's the thing about a T-shirt: the drawing is the product. Nobody buys a tee because the product page loads fast or the meta description is well written. They buy it because a melted purple blob with spiral eyes looks the way their Tuesday felt. That recognition is the whole sale.
And recognition is fragile. Fuzzy only works because every wobble is deliberate: how far the melt goes before it stops being funny, how much side-eye Duckling can give before he turns mean. Those are taste calls, made a hundred times per design, by someone who's been drawing since he could hold a crayon. That's design judgement, not a prompt.
There's a trust side too. This is a brand for a community that gets talked over a lot. Hand-drawn says a person made this, for you. A generated design, however polished, quietly says the opposite.
So the split is simple: the machine runs the shop, I make the thing people buy.
What this means if you draw for a living
Hand AI the business, keep the craft. The parts of creative work that used to stop people starting (the website, the admin, the SEO, the marketing grind) are now the cheap bit. That's not a threat to illustrators. It's the first time a lot of us can run our own thing.
Decide your line on purpose. "Should I use AI?" is the wrong question. Ask which part of the work your audience is actually paying for, and protect that. For me it's the drawing. For you it might be the concept, the voice or the eye.
Taste is the moat. Anyone can now ship a shop in a weekend. What they can't ship is your hand. It's the same lesson as writing a book in three days and building an app from a craft stall: the machine does the volume, a human decides what deserves to exist.
If you want to see how that split looks in practice, go and meet Duckling and Fuzzy at cutlerycrisis.com. Every line on every tee is drawn by hand. The rest, the machine handled.