Human craft in branding: why visible care beats AI perfection
, 6 min read
Quick answer
How does a brand show human craft when AI-perfect output is everywhere?
A brand shows human craft by making choices a generator would not make, and by leaving visible proof of them where customers touch the brand: on the pack, in the type, in the images and in the small details. When any team can produce a flawless render in seconds, polish proves nothing. IW Agency treats craft as evidence in three parts: the mark, the choice and the care. Research on the handmade effect (Fuchs, Schreier & van Osselaer, 2015) shows people read effort as love, and a copied style cannot fake real provenance.
How does a brand show human craft when AI perfection is everywhere?
A brand shows human craft by making choices a generator would not make, and by leaving visible proof of them where customers touch the brand: on the pack, in the type, in the images and in the small details. Polish no longer proves anything, because any team can now produce a flawless render in seconds.
At IW Agency, we think craft now works as evidence that someone cared enough to decide. In our view, it shows in three places: the mark, the choice and the care.
The mark is a visible trace of making. A choice departs on purpose from what the category expects. Care sits in the details nobody asked you to look at.
Morakniv shows what this looks like as a position rather than a style. The case starts from a stance: in a world increasingly willing to compromise on quality, Morakniv stands for something different, with its roots in a Swedish crafting village.
Why does human craft matter more when AI makes everything look perfect?
Human craft matters more because people read effort as love, and AI output shows no effort. In research on handmade products, consumers preferred items labelled handmade. Part of the reason was that they saw the products as containing the maker's love, and the effect was strongest when they bought gifts for people close to them (Fuchs, Schreier & van Osselaer, Journal of Marketing, 2015).
The study came out before generative AI, and that makes it more useful today. It shows what people were responding to all along. The object mattered less than the sense that a person stood behind it, and generated content removes exactly that signal.
Brands also cannot assume that people trust machine output. A Gartner survey found that 53 per cent of consumers distrust AI-powered search results (Gartner, 2025). As more of what people see is made by machines, we think they look harder for a human who answers for it.
Sameness is the other cost. Generative models produce the most probable result, and the most probable result is what everyone else already has. So typography gains weight in the AI era, and so do a brand's distinctive brand assets.
Conclusion: Perfection is now the baseline, and nobody remembers the baseline.
What does deliberate imperfection look like on pack and in comms?
Deliberate imperfection is a designed departure with a reason behind it, never sloppiness passed off as character. Each of the three parts takes a concrete form:
The mark. Hand-drawn lettering, illustration with real pen pressure, and photography of actual production instead of generated product shots. The mark has to come from the brand's own making.
The choice. One element that breaks the category's expected grammar, such as a colour nobody else uses, an off-grid layout or copy with a real voice. If you can't say why you made it, drop it.
The care. Detail found on the second look: inside the lid, under the flap, in the small print and in the design system that keeps it consistent across every application.
Packaging is where this pays most. A pack is picked up, turned over and kept in the kitchen. The Lurpak brief was to help a brand built on 120 years of craft heritage evolve and become future ready.
The same logic applies to voice and image in comms. Our work with Google at Almedalen began with a wish to offer a more human perspective on technology, at a time when AI increasingly shaped public debate.
Can't AI just fake the handmade look?
AI can fake the look of handmade, and that is the strongest argument against our position. Grain, wobble, brush texture and "hand-drawn" fonts are one prompt away. If imperfection is only a style, it will be copied within a season and stop meaning anything.
We accept the premise and reject the conclusion. What resists copying is a body of consistent, specific choices tied to something true. The handmade effect rests on the belief that a person really stood behind the object. Fake wobble borrows that belief without earning it, and one exposed "artisan" image can undo a great deal of credibility.
The test is therefore provenance rather than aesthetics. Can you show who made it, where and how? Is the mark specific to your brand, or could it sit on any competitor's pack?
None of this means refusing AI. Keep it where it serves production, as we describe in our piece on AI-generated images and copy without losing distinction. The proof of care should stay human. The EU AI Act's labelling rules are one more reason not to build a craft story on generated images.
Conclusion: A style can be copied. A provenance cannot.
What does this mean for your brand?
Audit where your brand is touched. List the pack, the site, the social feed and the store. Note where a human decision is visible and where only polish shows.
Choose one mark only your brand has. That can be your own lettering, photography of real production or real people. Make it specific enough that no competitor could borrow it.
Defend one deliberate departure. Pick the category convention you break and write down why. Without a stated reason, a choice reads as a mistake.
Put care into the second look. Budget for the details people find later, such as the inside of the pack, the small print and the unboxing. That is where the emotional connection gets built.
Keep generated assets away from your proof points. Use AI for volume and variation. Have people make the images and claims that carry your craft story.
Frequently asked questions
Is human craft only relevant for premium or heritage brands?
No. Everyday brands can show the mark, the choice and the care too, and in crowded categories it often matters more. On a supermarket shelf, a specific trace of making is one of the few things a private label cannot easily copy.
Does deliberate imperfection hurt brand consistency?
No, as long as it is designed into the system. Consistency means repeating recognisable choices, not identical pixels. A good design system specifies where the mark appears, how much variation is allowed and where the rules hold firm.
Should we tell customers that something is made by people?
Show it before you say it. A "made by humans" badge is a claim, while visible craft is proof. If you do say it, make it true and specific: who made it, where and how.
Can we still use AI if human craft is our proof?
Yes. AI is useful for production, adaptation and testing. Keep the assets that carry your proof of care human and the rest can scale. For the broader question of what AI changes in design work, see will AI replace design agencies.
How do you know your brand's craft is showing?
Look at your brand where customers touch it and ask whether a person's decision is visible. If only polish shows, there is no proof that anyone cared. To talk it through, write to us at hello@iwagency.com.
Sources
IW Agency is a Stockholm brand agency working across strategy, design, communication, digital experience and innovation. Founded in 1992 as Identity Works.
Facing the same question?
Tell us what you are working on, and we will get back to you.
Read next
Typography in the AI era: why type becomes a distinctive asset when everything looks generated
Typography matters more for brand recognition when AI makes content look the same. What research shows, the strongest objection, and when custom type pays.
Design, , 7 min read
Will AI replace design agencies? What a brand is worth when everyone can design
AI will replace design production, not the judgement that makes a brand worth choosing. Why distinction and meaning gain value when everyone can design.
Design, , 6 min read
Labelling AI-generated content: what the EU AI Act requires of brands
Do you have to label AI images and text in marketing? What Article 50 of the EU AI Act requires since 2 August 2026, who must label it and what is exempt.
Comms, , 7 min read