What a hundred-year-old silver tray can teach us about doing creative work in the age of AI.
By now, if you have hired an agency in the last year, you have probably opened a deliverable and known, almost instantly, what made it. You know it the second you open the link.
The design has a certain beigeness. The hover states on the prototype feel familiar. The colors and stylings carry the faint accent of a Claude artifact, or Figma Make, or v0, or whichever tool happened to be open that afternoon. You cannot always name it, but you can feel it. And it looks more or less like what you were shown by the last three firms.
I’m not here to complain about it. A single screen that used to take a designer two to four hours can now be generated in two to four minutes. Tools like v0 turn a sentence into production-ready React. Figma will tell you most designers already feel faster with this stuff and they’re right, they are faster. The interactive prototype, once the expensive proof that a firm had done the work, is now something a product manager can spin up before the first intro call.
Nothing is wrong with this. We use these tools at Bottle Rocket every day and we will keep adopting whatever the newest and most capable version turns out to be. That is not in question. But I think a lot of clients and agencies are wondering, when the artifact itself becomes nearly free to produce, where does the value of the work actually live and how do you make that value visible to the person on the other side of the table?
This artifact’s got legs
A good-looking prototype was never the problem, after all, they’re meant to test ideas, new journeys, new concepts, not final design systems. The real problem is that a prototype built on three weeks of deep research, collaborative workshops, and iterative testing looks, visually, exactly the same as an artifact built with one lazy Claude prompt.
And then it leaves the room. The design gets forwarded, dropped into a channel, opened by someone who was not on any of the calls. And when it leaves the room, it leaves behind everything that gave it meaning - The research, the customer interviews, the tradeoffs weighed and rejected, the reason this button is here and not there, the forty versions that came before this one. All of that stays in the room. The artifact arrives naked, stripped of the thinking that produced it. How is that CMO looking at two beige prototypes supposed to know which is which?
In this moment, production may be getting more commoditized but the judgment behind it is invisible unless you drag it into the light and point at it.
We’ve run this experiment before
In the second half of the nineteenth century, industrial production flooded the market with goods that were cheap, uniform, and technically competent. A group of designers and thinkers led by William Morris pushed back, starting the movement we now call Arts and Crafts. It is usually remembered for its wallpaper and furniture, but it was really about dignity, about the meaning behind the act of producing the work, and about what happens to both when a machine can produce the thing faster and cheaper than before.
Around the turn of the century, the Gorham Manufacturing Company introduced a line of silver called Martelé, French for "hammered." The pieces were covered in shallow dimples, the visible record of every hammer blow that raised them. Now, Gorham's silversmiths had the technology and the skill to buff those pitchers and trays to a flawless machine-smooth finish. They chose not to. The hammer marks were left on purpose, a deliberate signal that human hands and human decisions had shaped the metal.
In 1900, a Martelé display took the Grand Prix at the Paris International Exposition, beating the polished competition, Tiffany included. A century later the surviving pieces sell at auction for tens of thousands of dollars. The evidence of the human hand was not the thing you tolerated, it was the thing you paid for.
The hammer marks are not a defect. They are the point.
It is worth being precise about Morris here, because the lazy version of this story turns him into a man who hated machines, and he was not. He used new printing and textile technology in his own workshops. His objection was never to tools. It was to work that hollowed out human judgment and left nothing but the surface. That distinction is exactly the one our industry needs right now. The question is not whether to use AI. The question is whether the thinking survives the tool.
The strategy is the hard part
The research is the hard part. The pulling apart of the shitty data is the hard part. Understanding a client's actual customers, their actual constraints, the messy realities of their business, and then making a thousand small judgment calls about what matters and what does not – This is the work. Morris understood this about his own craft. In making any real object, he argued, the maker faces countless decisions and it is those decisions that determine whether the result is any good. The polished artifact was just the last, easiest mile, and now that mile is nearly free.
So the thinking is the product. It always has been. AI has simply made that undeniable by removing the one thing we used to hide behind, which was the labor of production. When anyone can generate a plausible surface, the surface stops being proof of anything. What remains is the reasoning, and reasoning has to be shown, not assumed.
So what are the hammer marks in our work?
If the hammer marks are the visible evidence of human judgment and good, hard work, then the practical question for anyone doing this work is, what do our hammer marks look like? A few to consider:
The “why” rides along with the deliverable. Annotate the decisions. Attach the reasoning to the artifact so it survives being forwarded to someone who was never on a call. A screen with a note explaining why it is built this way is worth more than a prettier screen with no explanation.
Show the cutting-room floor. The single strongest signal of real thought is the work you did not ship. The directions you tried and rejected and the reason you rejected them, tell a client more about your judgment than the final direction ever could. The forty versions are the proof.
Show the prompts. Do not hide the tool. Show the human steering it. There is nothing embarrassing about using AI to get somewhere fast. What matters is the intent behind the prompting, the iteration, the taste applied to the output. That is craft and it is worth making it legible.
Ground the work in things no tool has. Claude can produce a beautiful default. It cannot know your client's specific customer, sitting in a specific situation, with a specific frustration that surfaced in a real interview. Specificity is the thing the average aesthetic cannot fake and it is where human work still clearly wins.
Have a point of view. These tools reach, by design, for the safe median. The most human thing you can do is make a particular, opinionated, defensible choice instead of the average one. A deliverable with a spine reads as human because it is.
A challenge to the industry and to ourselves
So this is partly a call to the people who buy this work to change what you ask for. Do not grade on polish, because polish is cheap and universal. Ask for the thought behind it. Ask what was considered and thrown away. Ask to see the prompts. Reward the reasoning and you will get more of it.
But it would be too easy to make this only a message to clients. It is a challenge to us as well. Morris's deepest worry was never really about the objects. It was about the makers and what happens to a person's work when it is reduced to operating a machine. The real risk of this moment is not that the output gets worse. It is that the people doing the work get quietly demoted from thinkers to prompt-runners and that the satisfying, human, judgment-heavy part of the craft gets automated away along with the tedious part. Keeping our people in the judgment, not just at the keyboard, is the version of this we owe ourselves.
Where this goes
I think we are going to see a return to something like the Arts and Crafts sentiment, this time as a reaction to AI industrialization. One WIRED essay has already named it the "William Morris effect," the idea that a leap in technology does not kill our appreciation for craft but sharpens it, raising the premium on demonstrable human judgment, on provenance, on intention you can actually point to.
The tools are not going anywhere, and neither are we. We will keep using them. And we will keep leaving our hammer marks on the work - The visible evidence that people thought hard about your problem and made deliberate choices in service of it.
Start asking every partner you work with to show you theirs.