Why AI Can't Commoditize Original Jewelry
Independent makers have been bracing for agentic commerce, the shift to AI assistants that find a product, size it up, and buy it on a customer's behalf with no human ever clicking. Earlier this year, Kevin Indig published a serious readiness playbook for it in Growth Memo: a checklist for whether those assistants can find your store and complete a purchase. I ran my own jewelry store straight through it.
I failed the field that supposedly matters most.
My store has no GTIN. A GTIN is the barcode-style ID that lets a shopping bot recognize the same exact product across a hundred different stores. No GTIN, and by the checklist's logic, you are invisible to the machines.
I have been making one-of-a-kind gemstone jewelry for 18 years at andreali.com. Every piece I make exists once. When it sells, it is gone. So of course I have no GTIN. There is no identical item on another store to match mine against. My own readiness report said it plainly: skip the GTIN, it does not apply to work like this. And the moment I saw that "failure," I stopped worrying about AI competition entirely.
Here is what the panic is missing, and it reaches a lot further than one-of-a-kind. The thing the checklist marks as your biggest gap is your deepest moat. And the moat is not really "you only made one." It is originality. That protects far more of us than the panic admits.
Key takeaway: AI shopping agents work by finding a cheaper copy of the same product across stores. When your work is original, there is no identical copy at another store to find, so there is nothing to undercut. One-of-a-kind is the purest version of this, but a distinctive design made in a small run is protected too. What AI actually commoditizes is generic work, not originality. You are not behind. You are playing a different board, one that is far better defended.
What AI actually made free
Something shifted this year, and it is bigger than jewelry. AI can now produce a competent version of almost anything that has already been done. Competent product photos. Competent descriptions. Competent designs assembled from patterns in everything that already exists. The cost of "good enough" fell to nearly zero.
But "good enough" is exactly where AI stops. It cannot make something that has never existed before. It cannot originate. It can only remix what it has already seen.
So the value moved. It moved off of knowledge, which is now free and everywhere, and onto the two things AI cannot hand anyone: taste, meaning knowing what is worth making, and execution, meaning actually making it. For a jewelry designer, those two words have a physical form. Your taste is your design eye. Your execution is the piece in your hands. AI can describe both. It cannot originate either. That is true whether you make a piece once or make your own original design twenty times.
How AI shopping agents actually work, and where they break
To see why original work is protected, you have to see how an AI shopping agent makes a decision.
These agents run on comparison. In plain terms, they find the same product, or something close to it, across many stores, then recommend the cheapest or best-reviewed one. Scan, compare, rank, done. That is the whole job.
For mass-produced products this works perfectly. Silver hoop earrings? An agent can line up 200 nearly identical pairs in seconds and sort them by price. The lowest price wins, and everyone else disappears.
But when a design is genuinely original, there is nothing to compare it against at another store. The agent cannot find a cheaper version, because no other store carries it. It cannot weigh the value against a benchmark, because there is no benchmark. The comparison engine, the thing that commoditizes everyone else, simply has no work to do on your original work.
| What the agent does | Generic product brand | Original brand (one-off or small runs) |
|---|---|---|
| Price matching | Finds the identical item across 10 stores and picks the lowest price | No identical item exists at another store, so there is no cheaper version to recommend |
| Brand loyalty | AI strips away the convenience that built loyalty, so shoppers switch for a few dollars | The search for something distinctive is an emotional pull AI cannot satisfy with a substitute |
| Design | AI-generated designs start a race to the bottom on price | A recognizable design becomes more valuable precisely because it cannot be mass-produced |
The rule underneath the table is simple. If your work looks like something anyone else makes, AI will find the cheaper one. If your work is distinctly yours, AI has nothing to compare it to.
This is not only my read. Indig's own playbook draws the same line: commodity products compete on price in one lane, and differentiated products get recommended as the best in another. For work that exists only once, there is no price lane to lose in the first place.
What if you make more than one?
Most independent makers are not one-of-a-kind studios. You design something original, then you make it more than once. So where do you sit? On a spectrum, and the honest version matters more than a blanket "you are safe."
| Where you sit | How protected | Your real exposure | The move |
|---|---|---|---|
| One-of-a-kind (the strongest case) | Fully outside the comparison game. Nothing to benchmark, nothing to undercut. | None on price. The piece exists once, nowhere else. | Keep making the thing that exists once. |
| Original design, small runs (most makers) | Protected from the price race. The identical piece is not on a competitor's site to be found cheaper. | If the design drifts toward generic, an agent can offer a shopper a close-enough substitute. | Stay unmistakably yours, and be the name the agent knows. |
| Generic designs (the exposed group) | Exposed. Lands in a comparison set like everyone else's work. | The full race to the bottom on price. | Get more distinctive, or get compared away. |
This is not about handmade versus machine. A handmade piece that copies a common template still lands in a comparison set. Distinctiveness is the line, not the tool.
There is one more thing that matters more for a small-run maker than for a pure one-of-a-kind studio. Because you cannot lean on absolute scarcity, being the maker an agent knows by name carries more of the load. Distinctiveness keeps you out of the price war. Being recommended by name is what puts you in the conversation at all. Both are within your control, and both are the real work.
Being un-copyable is not the same as being found
Here is where most of the reassurance stops, and where the real work starts.
Your original work is protected from commoditization. That is true, and it is yours. But an AI agent cannot recommend a piece it never sees. Being hard to copy does not make you easy to find. Those are two different games played on two different fields.
Field one is your craft: the design itself, original, hard to substitute. You already own this.
Field two is discovery: whether an AI assistant, when a real person asks it for "a unique, handmade gemstone necklace in soft blues," actually knows your name and puts you forward. Most makers own the first field and have never touched the second.
And the second field matters more every month. It is easy to assume that because you do not shop with AI, your buyers do not either. The data says otherwise. 74% of households earning over $100,000 already use AI tools, compared with 53% of households under $50,000 (Menlo Ventures; Epoch AI and Ipsos). The people most likely to buy a $400 pendant are the people most likely to be asking an AI what to buy. The fact that you may not shop this way is not evidence that your customer does not. They are not you.
So the goal is narrow and specific. Be original enough that AI cannot compare you away, and legible enough that it can still find you and hand you to the right person.
Why this window is open right now
There is a reason to move on this now rather than someday.
We have seen this shape before. When search engine optimization was new, it was a level field. The playbook was unknown and cheap to run, so a single sharp person could out-maneuver a slow, giant company. Then it stopped being a thinking game and became a spending game, all budget and volume and scale, and the giants took it. The solo operator did not get worse at it. The board changed to one only money could play.
What is different this time is that the thing that closed the old window is the thing AI erodes. The advantage the giants used to win search was cheap execution at scale, which used to require a whole team. A solo maker with taste and the right tools can now produce that level of work alone. The field is level again, for now, because nobody has the playbook yet and the old budget advantage has thinned.
It will not stay level. Big companies will figure this out and arrive with their distribution and their budgets, and they will take the broad, generic layer the way they took broad search. What they will never do is serve the long tail of individual makers with real care, because it is too small and too particular to be worth their while. That is the opening. Not to beat them on their stage, but to become the trusted, specific name in a corner they will never bother to enter, and to get there while getting there is still cheap.
The real threat is not what you think
Here is the honest part, and it is not about bots.
If your work is genuinely original, AI agents cannot commoditize you. The real pressure coming for original makers is different. As AI reshapes industries, some of your customers may have less to spend. The person who bought a $400 pendant without a second thought might tighten up because their own work was touched by automation.
That does not put you on defense. It moves you toward the aspirational. When people buy fewer things, they buy fewer things that hold real value. Your work needs to be one of those things. Not a nice-to-have, but the piece someone chooses precisely because it is real, made by a specific human, holds its worth over time, and cannot be reproduced. Scarcity you can prove becomes an asset the moment discretionary money gets careful.
Three moves for makers who do original work
Being original is the moat. These three moves make the moat visible and load-bearing, whether you make one piece or a small run.
1. Make your original work legible to machines. Your product data needs to be clean enough for an AI agent to understand you, and your work needs to be distinctive enough that it can never compare you away. I call this Agent Intelligence Optimization, or AIO. In practice it means your descriptions, your structured product tags (the behind-the-scenes labels that tell an AI what a page actually is), and your specifics, the materials, the technique, the gemstone origin, the dimensions, the inspiration, are all detailed and exact. The richer your data, the more accurately an agent can match you to the right buyer, and the more your distinctiveness reads through. You are teaching the machine to respect what makes you yours.
The descriptions you can write yourself. The structured-data layer underneath, the part that turns your words into fields an agent can actually read, is the piece most makers hand to someone who does this work. Either path is fine. What matters is that it gets done, not that you become a technician to do it.
2. Tell the story, because the loyalty is to the maker. For an original brand the relationship is not with the transaction. It is with you, and with the story behind the design. Instead of marketing to ten thousand people, use these tools to tell the story of the work to the right hundred. The inspiration behind the stone. The choice that makes it recognizably yours. That story is not marketing copy. It is provenance, the documented history of the work, and provenance builds a loyalty AI cannot disrupt, because the loyalty is not to a brand that can be undercut. It is to a person.
3. Position the work as an investment, not an impulse. In a market flooding with AI-generated products, original handmade jewelry becomes what economists call a Veblen good: an object that grows more desirable precisely because it is scarce and hard to produce (Thorstein Veblen named this in 1899). When "good enough" is free and everywhere, the premium moves to "irreplaceable." Your hand work, your eye, your sourcing, your years of craft are not only your creative identity. They are a financial moat.
The makers who thrive
The designers who will do well in this next stretch are not the ones hiding from AI. They are the ones who lean all the way into what makes them original, and then make that visible to both the humans and the machines.
You already make the thing AI cannot. The only question left is whether it can find you.
Frequently asked questions
Do AI shopping agents ignore original or handmade products?
Not ignore, but they cannot commoditize original work. An agent's main trick is finding a cheaper identical item at another store, and there is no identical item to find. As long as your product data is detailed and specific, an agent can still surface your work to the right buyer. It just cannot use that data to undercut you.
I make small runs of my designs, not one-offs. Am I protected too?
Yes, as long as the design is distinctly yours. The identical piece is not sitting on a competitor's site to be found cheaper, so you stay out of the price race. Your exposure is different from a one-of-a-kind studio: if a design drifts toward generic, an agent can offer a shopper a close-enough substitute. The more recognizable your work, and the more you are the name an agent knows, the safer you are.
Do I need a GTIN or a barcode to be found by AI?
No. A GTIN identifies the same product sold in many places, which is the opposite of what you make. Skip it. Put your effort into rich, specific descriptions and structured product data so an agent understands what your work is and who it is for.
Will AI-generated jewelry put independent designers out of business?
It will crowd the generic, mass-produced end hard. It cannot touch genuinely original work, because AI remixes what already exists and cannot originate. The designers at risk are the ones whose work already looks like everything else. The more distinctive your work, the more defended you are.
Sources
- Agentic-commerce readiness playbook: Kevin Indig, Growth Memo, "How do you compete in Agentic Commerce?" (January 2026). Its own positioning logic splits commodity products (price lane) from differentiated products (recommended-as-best lane).
- Product Data Readiness Report for andreali.com: Red Pin Geek, February 2026 (the live audit referenced in the opening; it found no product carries a GTIN and recommends skipping GTIN for handmade one-of-a-kind work).
- AI adoption by income: Menlo Ventures (2025); Epoch AI and Ipsos (2026).
- Veblen goods: Thorstein Veblen, The Theory of the Leisure Class (1899).
- Comparison-engine mechanics and Agent Intelligence Optimization (AIO) are the author's framework, developed through a live audit of her own store and client work, used with permission.
Continue Learning
- The Foundation Gap: Why AI Search Skips Independent Jewelers: the second field in full, why beautiful, hard-won sites still get passed over, and the foundation that changes it.
- Jewelry SEO in 2026: the complete guide to ranking on Google and getting found by AI: the whole playbook and checklist, if you want the full map.
- The full AI visibility audit of my own store (case study): I test every method on andreali.com before a client ever sees it. Here is the receipt.
- Score your own store on the free AI Visibility Score: see whether an AI assistant can actually find and describe your work today, the same read I ran on my own store.
- Subscribe to the Red Pin Geek Substack (free): weekly, inside-view notes on building AI visibility as a maker. The Premium tier unlocks the worked tools and templates.
- Work with me on your specific store (AI Visibility Snapshot $97 / Full AI Visibility Audit $597 / Full Audit + Strategy Call $997): I run this exact read on your store and tell you what an AI sees when it looks at you.
Want Red Pin Geek in Your AI Answers?
Google now lets you choose preferred sources that get labeled in AI Overviews and AI Mode. If this content helps you navigate AI visibility for your jewelry business, add us and you'll spot our links faster next time Google answers your questions.
Add Red Pin Geek as a Preferred SourceTakes one click. Works in Google Search, AI Overviews, and AI Mode.
