AI Knows You Exist, But Can It Describe You?

Key Takeaway

Key takeaway: Being found by AI and being understood by AI are two different things. AI can surface your name and still describe almost nothing true or distinctive about you, repeating maybe 20% of your real story and quietly missing the 80% that makes someone choose you. That missing 80% is your blind spot. You cannot see it because from where you sit you know the whole story; the shopper, through AI, only gets the sliver.

There is a beat in the AI search conversation that almost nobody has named yet. The convergence piece I published last week walked through five independent experts arriving at the same warning: brands built on vibe alone are vanishing from AI answers. The whole conversation has been about getting found.

This piece is about the gap on the other side.

Being mentioned in an AI answer is not the same as being understood by the AI. A thin, generic, or subtly wrong description is what loses the sale even when you ARE in the answer. The shopper choosing between you and three other makers does not pick the one AI named. They pick the one AI could describe richly. The thinly-described maker gets skipped past, even after they were surfaced.

Here is the part that makes this hard to see on your own work: from where you sit, you already know your whole story. The shopper, through AI, only gets the sliver. You have no way to see what they are seeing.

I want you to feel this gap on your own brand in the next two minutes. Then I will show you what I found on mine.

Try this on your own brand right now

Open ChatGPT (or Perplexity, or Google AI Overview, whichever you have handy). Paste this prompt, with your real brand name in the brackets:

Describe [YOUR BRAND NAME] in 3 to 4 sentences. What kind of jewelry do they make? Who is the designer? What is special or different about them as a brand? Why might someone choose them over another independent jewelry maker?

Read what comes back. Read it carefully.

Then ask yourself one honest question: would a stranger reading just that description know what is actually special about your work? Would they know the parts of your story that drive the choice? The training that shaped your hand. The reason your brand exists in the first place. The specific thing about how you make and what you make that means more than the architectural facts.

If the answer is no, you have just found your blind spot. And it is the same blind spot a shopper sees every time they ask an AI for a recommendation that should include you.

Sit in the result for a moment

When I ran this prompt on my own brand, Andrea Li Designs, ChatGPT came back with a sketch that included my name and called my work "story-driven" with "symbolism, texture, color." Reasonable enough on a first read. But it missed Denver. It missed gemstones and pearls. It missed sculptural and intricate wirework. It missed one-of-a-kind discipline. It missed vintage incorporation. It missed the cold-connection technique that exists because heat damages stones. It missed eighteen years.

And then it added something I had not said and would never say about myself: it called me someone with "a marketer's understanding of how personal style becomes part of someone's identity." I am a jewelry maker, not a marketer. ChatGPT had no truth to fill in, so it invented a flattering-sounding one.

That is the first failure mode of the Blind Spot. The description is thin, the maker's real story is mostly missing, and what little context AI does provide is sometimes made up to plug the holes.

Failure mode A on Andrea Li Designs

ChatGPT's response when asked to describe Andrea Li Designs in 3 to 4 sentences. The architectural facts that drive purchase are missing; ChatGPT invented a flattering substitute to fill the gap.

What ChatGPT captured

~20%

• The brand name
• "Original" and "story-driven" as vibe words
• "Symbolism, texture, color" as adjectives

What ChatGPT missed

~80%

• Denver location
• Sculptural gemstone and pearl jewelry
• One-of-a-kind discipline
• Intricate wirework technique
• Cold-connection technique (heat damages stones)
• Vintage and antique reconfiguration
• Recycled-gold material discipline
• The 18 years on the bench

Plus the hallucination layer: ChatGPT also invented a fact it could not have known and that is not true. It called Andrea someone with "a marketer's understanding of how personal style becomes part of someone's identity." Andrea is a jewelry maker, not a marketer. The AI made up a flattering-sounding substitute for the real story it could not see.

Source: Andrea Li, Red Pin Geek multi-AI brand-description scan, 2026-06-09. ChatGPT response on the prompt "Describe Andrea Li Designs in 3 to 4 sentences..."

Now here is the part that took me by surprise when I ran the same prompt across three different AI tools on the same day. Perplexity came back with a rich description. It captured the Denver location, intricate wirework, sculptural gemstone and pearl jewelry, "intentional color theory," "wearable sculpture," "personal talisman" framing, and even the client-experience language about feeling powerful and seen when wearing my work. Google AI Overview came back richer still. It captured all of that plus the unconventional materials I use (laser-cut wood, plexiglass alongside the gemstones), the anti-trend architectural framing, the "artistic metamorphosis" of antique reconfiguration, and the specific buyer persona my work serves.

Two of the three AIs described my brand the way I would describe my brand to a serious collector. Almost all of the load-bearing brand identity made it through. The blind spot is not uniform across AI tools.

The thin description from ChatGPT is the specific failure I want you to look at hard. Because ChatGPT is the AI most shoppers are actually using to shop, the gap there is the gap that costs the sale, even when Perplexity and Google AI Overview see your brand clearly. The shopper who asks ChatGPT for a recommendation never sees the rich Perplexity description. They see the thin one.

Why does ChatGPT miss what Perplexity and Google catch? Because Perplexity and Google do live web retrieval at query time. They reach out to your site, your published content, and the third-party pages that reference your brand. ChatGPT runs primarily on training-cutoff knowledge with limited live retrieval. The brand-identity foundation work I did on andreali.com lives on my site, which Perplexity and Google can read at query time. But the same work has not yet propagated into ChatGPT's training data, because to get there it needs to be cited and referenced in third-party content that AI training cycles ingest. My own site cannot cite itself into training data. The brand-identity layer needs a citation network around it to reach the training-cutoff AIs.

The two failure modes

The thin-description failure is the obvious one. There is a second failure mode I have watched happen, and it is subtler and more insidious because the brand owner does not see it.

AI captures the architectural facts about a brand: designer name, location, technique, material. But it completely misses the load-bearing differentiation that drives purchase. A personal narrative behind why the brand exists. A specific sourcing story. The philosophical framing that makes the brand mean something. The description reads as "mostly right" on a first scan, which is exactly why the brand owner does not see the gap. But the parts AI missed are the parts that actually drive the choice.

A shopper deciding between three makers does not choose the one whose designer name is captured. They choose the one whose reason-for-being lands. If AI cannot describe the reason-for-being, the maker gets passed over even with their architectural facts intact.

Failure mode	What AI captures	What AI misses	Why it costs the sale
A. Thin description	Roughly 20% of the brand: name, a vibe word or two. Sometimes hallucinates plausible-sounding details to fill the gap.	70-80% of the real brand: location, materials, technique, designer identity, founder story, anything specific.	Shopper cannot tell what the brand is or why it matters from the description alone.
B. Architecturally rich but soul-missing	60-70% of the architectural facts: designer name, location, technique category, material category, brand-positioning summary.	The load-bearing differentiation that drives purchase: the personal narrative behind why the brand exists, the specific sourcing story, the philosophical framing.	Shopper sees "mostly right" and moves on. Brand owner reads it and thinks "well, that is mostly right." Nobody sees the gap.

Failure mode B is the more dangerous one. It hides in plain sight. The brand owner reads the description and thinks the brand is described correctly. The shopper reads the description and chooses someone else, because the parts that would have made them choose THIS maker were not in the answer. Nobody in the loop sees the gap clearly.

The asymmetric proof underneath the framing

I ran the same prompt on Bohemi, the studio I helped rebuild as a client. Heather Ng's brand. Boulder, Colorado. Alternative engagement rings, custom heirloom redesigns, recycled metals, fair-trade stones.

The descriptions came back rich. Across ChatGPT, Perplexity, and Google AI Overview, the AI named Heather, named Boulder, named the alternative engagement ring positioning, named the recycled-metals discipline, named the fair-trade sourcing, named the philosophical framing she has built around the work. Perplexity captured her brand's own poetic phrasing ("talismans made from earth and ore") which she uses on her own site. The descriptions were not perfect. But the load-bearing differentiation was there.

Same designer category. Same prompt. Sharply different result. Especially in ChatGPT. Why?

When I rebuilt Heather's site, we did the page-level foundation work AND the brand-identity layer work. The facts under the atmosphere. The clear product taxonomy. The explicit philosophy. The named sourcing standards. The page where her story lives in plain English. That work landed in retrieval AIs the same way it has on my own site. Perplexity and Google AI Overview describe Heather's brand richly because the work is documented on bohemi.com where they can read it at query time.

But the part that explains the ChatGPT difference is the next layer up. I did not only rebuild Heather's site. I also wrote a case study about Bohemi on my own site, walking through the structural work, the results, and the methodology. That case study is now public on redpingeek.com, where it gets indexed, gets cited, and gets referenced by other content. As of 2026-06-09, Google AI Overview cites it as source [2] in its description of Heather's brand. The case study is in the citation network around Bohemi in a way that nothing equivalent exists for Andrea Li Designs yet.

That is the asymmetric proof. The page-level and brand-identity foundation work landed where retrieval AIs read at query time on both brands. Bohemi's training-cutoff AI advantage is not the site rebuild alone. It is the third-party citation network around the brand. The site work made the brand legible. The case study and the citations made the brand legible IN the training data that ChatGPT ingests.

The thin ChatGPT description of my own brand is not "the methodology does not work on Andrea Li Designs." The methodology absolutely works at the site level and the retrieval-AI level. What the scan revealed is that the description gap on training-cutoff AIs is a citation-network problem, not a site-content problem. I found that by running my own brand through my own scan, the same way I found the page-level gap on my store three years ago. The Brand Description Gap tool, the one I am dropping in this cascade, is the diagnostic for the citation-network layer. I am working on building the citation network around andreali.com publicly over the next few weeks. The before-and-after will be in a follow-up piece. You can watch the practitioner story add its next chapter in real time.

One sharpening that landed when I sat with the scan results. The "marketer's understanding" detail ChatGPT invented about Andrea Li Designs was not random. I have built a dense citation network around Red Pin Geek (my consulting brand) over the past few years. I have not yet built an equivalent citation network around Andrea Li Designs (my jewelry brand). Because both brands share a founder, the consulting brand's citation density is bleeding into ChatGPT's description of the maker brand, filling the description gap with the closest adjacent material it has. Same mechanism as the Bohemi case study feeding Google AI Overview's description of Heather, running in the opposite direction. The fix is identical: build a citation network specifically for the brand that needs description-layer signal. Heather has hers because I wrote it. I have to build mine for andreali.com in publications that name me as a maker, not a marketer.

The load-bearing proof point

Andrea's case study is now an AI-cited source for Bohemi

When Andrea asked Google AI Overview to describe Bohemi on 2026-06-09, the response included its standard citation list. Source 1 was Bohemi's own site. Source 2 was Andrea's case study on Red Pin Geek.

Google AI Overview citations, "Describe Bohemi..." query, 2026-06-09

[1] bohemi.com

[2] redpingeek.com/case-studies/bohemi-engagement-ring-authority

[3-12] additional bohemi.com pages + third-party directory listings

The foundation work feeds itself. The case study Andrea wrote about Bohemi, on her own site, is now the public-record source AI reaches for when someone asks about Heather's brand. The methodology Andrea has been teaching is now visible in how an AI describes the brand she rebuilt, on a search system shoppers use daily.

Verify this yourself: run the prompt "Describe Bohemi in 3 to 4 sentences..." on Google AI Overview. The citation has been present since at least 2026-06-09.

The single strongest piece of evidence

Here is the part that made me stop and screenshot.

When I asked Google AI Overview to describe Bohemi, it returned a rich description with proper citations underneath. Source [1] was bohemi.com. Source [2] was redpingeek.com.

The case study I wrote about Bohemi, on my own site, is now a cited source in Google AI Overview's description of Bohemi. My own content is feeding back into how AI describes my client's brand. The foundation I helped her build is now part of how the world's most-used search system understands her work.

You can verify this yourself. Run the same query. The citation has been there since at least 2026-06-09.

That is the methodology feeding itself in real time. The piece of content where I documented what I built for Heather is now the piece of content AI reaches for when someone asks who Heather is. The proof is not that I am claiming this works. The proof is that AI is doing it, on the public record, citing my work.

I want to be careful with how I frame this. It does not mean foundation work guarantees AI will describe you perfectly forever. AI systems change. Citations come and go. What it means is the foundation work made the brand legible enough that the system noticed and pointed to the documented proof of it. That happened because the foundation was built, not because something was wished into being.

Why this hits independent makers specifically

The makers who have already accepted AI matters, maybe earned a single citation, run the X-Ray on a page, and concluded they are fine are exactly the makers most exposed to failure mode B. They have done enough work to show up in the answer. They have not done enough work for the answer to describe them in a way that drives the choice.

This is the part of the conversation that has been missing from the industry. The convergence pieces I gathered in my last post all named the visibility problem. None of them named the description problem on the other side. Visibility is the gate. Description is what happens after you walk through it. If the description on the other side of the gate is thin or soul-missing, the gate did not actually open anything for you.

The fix is not "write more poetry on your product pages." The fix is making sure AI has the specific, true facts about your brand and the load-bearing reasons-for-being that drive purchase, where AI can read them, in language AI can repeat back to a shopper. That is the work the Vibe Translator (the tool I am building for the next cascade in this series) is designed to do. For now, this piece is about seeing the gap. The fix comes next.

If this is starting to feel like a lot, hear this

The methodology has layers. I keep finding new ones because I keep running my own brand through my own scans. That is the practitioner job, and it is genuinely large work. If you are reading this and feeling the weight of "great, another thing I need to learn and build on top of everything else I am already doing," I want you to know I see you. You are not behind. The methodology is big and you are one person at a bench.

Here is what you actually need to do this week, in plain terms.

1. Run the prompt above on your own brand. Five minutes. Read what comes back cold, as a stranger would.
2. If the description was thin or missed the parts that drive purchase, you found your blind spot. That is the whole exercise.
3. Decide whether to keep going yourself or hand it off.

Keep going yourself: subscribe to the Red Pin Geek Substack (free) for the proof letter, the upcoming Brand Description Gap walkthrough, and the rest of the Foundation Gap arc. Run the tool on your top pages when it drops. Apply the fixes one page at a time over a few weeks. The methodology is teachable and the tools are built specifically so you can DIY.

Hand it off: this is the work I do. The audit and implementation services I offer walk you through your specific store. I run the same three-layer scan I just walked you through and write up exactly what is missing in your specific brand's foundation, identity, and citation network. Snapshot is $97. Full audit is $597. Full audit plus a one-to-one strategy call is $997. The audit exists because the methodology became too much for one practitioner to do as a side project. I built the practice so you can stay at the bench.

Frequently asked questions

My AI description was mostly right. Is that good enough?

It depends on what was missing. If AI captured your architectural facts (name, location, materials, technique category) but missed the personal narrative behind why your brand exists, the specific sourcing story, or the philosophical framing that drives purchase, you are likely in failure mode B. The description reads as "mostly right" but the parts AI missed are the parts that make a shopper choose you. Run the Brand Description Gap tool to see which differentiators specifically are not making it through.

How is the Brand Description Gap tool different from the AI X-Ray?

The AI X-Ray reads a single product page. It tells you what an AI can extract from that one page's copy. The Brand Description Gap reads your whole brand identity. It tells you what AI says about you as a brand and what specific differentiators are missing from how AI describes you across the surface area of your site. Different layers of the same Foundation Gap.

Why does my brand get described better in Google AI Overview than in ChatGPT?

ChatGPT runs on training-cutoff knowledge with limited live retrieval. Perplexity and Google AI Overview do live web retrieval at query time. If your foundation is well-built and your story is documented somewhere AI can read at query time (your own site, your case studies, your published content), the retrieval-based AIs will return richer descriptions. ChatGPT will catch up over time as new training cycles fold in the documented content, but the retrieval AIs are where the better descriptions live first.

If I rebuild my foundation, will AI describe me perfectly?

No. The honest framing is that rebuilding the foundation makes your brand legible to AI, which makes accurate description possible. It does not guarantee any specific AI will describe you any specific way. AI systems change, retrieval changes, citation patterns shift. What you control is the legibility of the source material. The foundation work makes the rest possible without making it certain.

Sources

Continue Learning

• The Foundation Gap (pillar): the full thesis underneath this piece. What changed in how people find jewelry, why the playbook we were taught no longer compounds, and the three layers of the foundation in plain English.
• Why Your Jewelry Site Isn't Getting Found Online: the cascade anchor on the visibility side of the gap. Where this piece picks up the description side.
• Why a Wave of Experts Just Named the AI Visibility Gap: five independent practitioners arriving at the same conclusion in a six-week window. The convergence piece this cascade rides on.
• Run the Brand Description Gap on your own brand: the Premium Substack tool that walks you through the same scan I ran on andreali.com and Bohemi.
• See the AI X-Ray on a single page: the page-level diagnostic from Cascade 1. Pairs with the Brand Description Gap on the brand-identity level.
• Score your full store on the AI Visibility Score: the free 30-second store-wide diagnostic. Useful as a backstop after running the page-level and brand-level tools.
• Subscribe to the Red Pin Geek Substack (free): the proof letter that pairs with this blog, plus the next cascade in the Foundation Gap arc (the Vibe Translator, the actual fix).
• Work with me on your specific store (Snapshot $97 / Full Audit $597 / Full Audit + 1:1 call $997): if you want the same brand-description gap diagnosed at scale on your store with the warehouse-to-machine read written up for your specific catalog, the audit is the private read.

About Andrea

Andrea Li is the founder of Red Pin Geek, an AI visibility consultancy for independent jewelry designers. She has spent eighteen years in the jewelry industry, including running her own jewelry brand at andreali.com, where she documented the convergence between traditional SEO and AI citation eligibility. She was previously the Pinterest coach for a well-known year-long mastermind program in the jewelry industry, and now teaches and builds the machine-readable foundation layer that program (and most others) never covered. Red Pin Geek is the practitioner-led methodology she developed first on her own store, then refined through client engagements including Bohemi and Talisman Fine Jewelry. Andrea writes for the jewelry designer at the bench, not the SEO professional at the desk.

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 Source

Takes one click. Works in Google Search, AI Overviews, and AI Mode.