Using AI As A Vintage Seller

As sure as the sun rises, as sure as the seagulls steal snacks at the beach, as sure as the doorbell rings as soon as the baby is down for a nap, it’s a certainty that any new technology can be co-opted in a bad way. AI, in the form of chatbots, is no different. From students using them to write papers to fraudsters bilking people out of their life’s savings; it’s unsettling how it can be utilized to do bad things. But people using it in bad ways does not mean it’s not a useful tool. For example, a hatchet can be infamously used by someone to give their parents 40 whacks (back in 1892), and it can be used in a helpful way to split kindling for a cozy night around the fire pit. Chatbots are a tool. Using AI as a vintage seller means finding the good ways to use that tool, while avoiding the pitfalls.

The major flaws with chatbots are:

They are designed to give you an answer to your query. If they can’t find the actual info, they will happily hallucinate an answer. Some chatbots hallucinate more than others.
They are sycophantic. They are not friends, but they act like friends. This builds trust that the chatbot should never earn. Stay skeptical, my friends.
They only know what they can scrounge off the internet and sometimes the sources they find are suspect, so always verify what sources they got your answer from.
And just because you correct them once on some point does not mean they will ever remember it again, at least when you are using the free tiers.

A companion post to this one discussed using AI as a research tool. In this post, I’m going to share my experience using AI as a vintage seller to streamline and improve my listing process. There are things it is good at and things it stinks at. Your own experiences may be different than mine. Results may vary. We’d love you to leave comments, because this is going to be an active discussion for vintage sellers for the foreseeable future.

IMPORTANT NOTE: I only use the free tiers that are available on ChatGPT, Claude and Google AI. If you are a paid user, you may be getting better results than me.

The most important thing to know about using AI is that it can only work from what you tell it. To give your chatbot a fighting chance at giving you useful information, add as much detail to your query as you can, evaluate the answer and ask follow up questions. Chatbots allow you to have multi-turn conversations, where in one thread, you ask a number of questions. It’s called prompt chaining when you use an answer to form a new query, and it can be very helpful in getting what you need. As long as you stay in one thread, your chatbot will maintain context and remember previous answers and prompts. ChatGPT and Claude save your conversations so you can reopen them to review and ask more questions. Regular Google AI does not.

Using AI to help write listings

AI can write a listing for you based on the information you give it, but once you’ve seen the output a few times, you can pick out an AI written listing that hasn’t been edited and personalized by the seller at five paces. 90% of them start with “Step into…” or “Bring vintage charm…” and end with “Perfect for…” followed by three examples.

But, when you’re churning out a number of listings everyday, having AI highlight key points that will drive SEO gives your brain a break. Trying to get SEO right for every listing is like trying to solve the New York Times Connections puzzle all day long. It’s fun once, but once a day is plenty. AI can also alert you to applicable style aesthetics, like cottagecore and grandmillenial, that might never cross your mind. That said, AI generated listings lack your personality and your touch. It’s commonly acknowledged that chatbots are not good at writing things. They’re adequate, but lean heavily into flowery, predictable marketing talk. You can reduce that by asking it to be more editorial, for example. But your writing will always be much better than anything AI spits out.

Let’s see our AI tools-not-friends in action. Using this photo…

And this prompt (I pasted in the description from my actual listing, which is a dry, clinical snoozefest):

Can you write a listing and title for Etsy? 1970s Gold and brown crocheted sleeveless mini dress – size medium. Beyond adorable fit and flare handmade crocheted mini dress in gold and brown. Sleeveless fitted top with back zipper. Open crochet skirt that’s semi-full. Zipper is nylon. Worsted weight yarn. Note: it’s pinned to fit the dress form.

Material /acrylic yarn

Care / machine wash

ChatGPT offers this…

1970s Gold & Brown Crochet Mini Dress • Handmade Boho Sleeveless Fit & Flare • Vintage Open Knit Festival Dress • Size Medium

Description:
Beyond adorable 1970s handmade crocheted mini dress in a warm golden yellow and rich brown colorway. This charming fit-and-flare piece has that perfect vintage boho vibe—easy to style for festivals, summer days, or layering.

The fitted sleeveless bodice features a simple round neckline with contrasting brown trim, while the waist is defined with a wide brown band. The skirt flares out into a semi-full silhouette with an open crochet pattern and decorative striping near the hem. It fastens with a nylon back zipper for easy wear.

A true one-of-a-kind handmade piece with great texture and movement.

Details:
• Era: 1970s
• Style: Fit & flare mini dress
• Color: Gold / yellow with brown accents
• Sleeveless with round neckline
• Fitted bodice with back nylon zipper
• Open crochet semi-full skirt
• Handmade

Size:
Tagged size: Medium (please refer to measurements for best fit)

Measurements (laid flat, unstretched):
• Bust: 18″
• Waist: 14″ (at waistband)
• Length: 33″

Material:
Acrylic yarn (worsted weight)

Care:
Machine wash

Condition Notes:
• Excellent vintage condition
• Dress is pinned on the mannequin to show fit

Perfect for vintage lovers, cottagecore/boho styling, or anyone who appreciates unique handmade pieces with a 1970s feel.

What ChatGPT did well:

It tipped me off to using “boho” and “festival dress,” two trends I had not thought of.
It emphasized important words for SEO in both the title and listing.
It is less clinical than my description, which was deadly unimaginative and dull, but it’s bloated with marketing speak.
It provided a good base for me to work from to improve the listing, including improving SEO.
It asked me if I wanted tags to use.

What ChatGPT did poorly:

Etsy is now judging us if we use more than 14 words in a title. This is more than 14 words.
Although it did not start this particular listing with the dreaded “Step into the…” that it uses 90% of the time, it did include the equally dreaded “Perfect for…” followed by three examples.
It insists on repeating information given in the description as bullet points. If you ask it why, it says repetition is smart because some people skim listings and SEO likes it when key words are reinforced. It does concede that some of the points could be removed from the description and only be in the bullet points.
It made up measurements. I have to assume they’re based on common measurements for a women’s size medium, which was stated as the size in the prompt. They are not the correct measurements for this dress.
It decided it was in excellent condition. I rarely if ever use “excellent” as a condition descriptor, and I didn’t give it any info other than the photo to reach that decision.

It’s not a bad listing. It does a lot of things right. But it doesn’t have my personality in it. Is that important? You have to decide that.

Here’s what Claude did from the same prompt:

1970s Gold & Brown Crochet Mini Dress — Size Medium | Handmade Fit & Flare Sleeveless Dress | Vintage Crochet Shell Stitch Skirt

This dress is an absolute dream. Hand crocheted in warm golden yellow and rich chocolate brown, it’s everything the 1970s did right — cheerful color blocking, a cinched waist, and just the right amount of boho charm.

The fitted bodice is worked in a dense, structured stitch that gives beautiful definition through the torso. Below the wide brown waistband, the skirt opens into an airy shell-stitch pattern that moves beautifully and catches the light. Brown scalloped trim edges the neckline, armholes, and hem for a polished, intentional finish.

Details at a glance:

— Size: Medium

— Style: Sleeveless fit & flare mini dress

— Closure: Back nylon zipper

— Skirt: Open shell-stitch crochet, semi-full

— Yarn: Worsted weight acrylic

— Colors: Golden yellow & chocolate brown

Care:

Machine wash gentle, lay flat or hang to dry. Acrylic yarn holds up beautifully with easy care.

Condition note:

Photographed pinned to fit the dress form — please refer to measurements for accurate sizing. Overall condition is excellent for the era.

What Claude did well:

It also tipped me off to boho.
It also got the important SEO terms right up front.
It is also less clinical then my description.
It worked with the information it was given and did not fictionalize measurements.
After the post, it added comments that emphasized the importance of adding measurements, suggested different views to photograph and added a list of tags.

What Claude did poorly:

Again, for Etsy the title was too long.
It’s a little breathless and gushy. I’m selling the dress and even I wouldn’t call it “a dream.”
It too decided the condition was excellent, but added “for the era” as a kind of disclaimer.
It repeats details as bullet points. When asked if that was essential, Claude said absolutely not. “The only bullets worth keeping are things like measurements or care instructions — genuinely scannable data that doesn’t fit naturally into prose. Everything else reads better as a story.”
I asked Claude why it included the bullet points, and it said: “Good question — I shouldn’t have. I defaulted to a common Etsy listing format without thinking critically about whether it actually served the content. The paragraphs I wrote were already detailed and covered everything, so the bullet recap was just filler that made the listing longer without adding value.”

After exchanges like that, you have to remind yourself that AI may be friend-shaped, but it’s not a friend. It’s a tool.

So what will the actual listing look like? I’ll take some of the important points from the AI-generated descriptions, tone down the salesman-speak and add in some of my personality.

The dress is something that lends itself to flowery marketing speech. What will ChatGPT and Claude do with something that needs a more to the point description, like this listing for currency? I did not provide the photo to the chatbots because they didn’t need it to write the listing, but I’m sticking it in here because there is a lot of type and not a lot of photos in this blog post.

The description I gave the chatbots:

5 Vintage Red Seal $2 Bills – 1953 to 1963 Series – U.S. Note Circulated Legal Tender

Own a piece of American financial history with this lot of 5 vintage red seal $2 bills, issued by the U.S. Treasury and spanning five unique series from 1953 to 1963. These are United States Notes, not Federal Reserve Notes. Each bill in this set comes from a different series, offering a unique snapshot of U.S. Treasury leadership and mid-century economic history.

Included Series & Signatures:

1953 – George M. Humphrey (Secretary) & Ivy Baker Priest (Treasurer)

1953A – Robert B. Anderson (Secretary) & Ivy Baker Priest (Treasurer)

1953B – C. Douglas Dillon (Secretary) & Elizabeth Rudel Smith (Treasurer)

1953C – C. Douglas Dillon (Secretary) & Kathryn O’Hay Granahan (Treasurer)

1963 – C. Douglas Dillon (Secretary) & Kathryn O’Hay Granahan (Treasurer)

CONDITION Circulated bills in mixed condition: 2 bills are crisp and clean; 3 bills show minor wear, including creasing, tiny corner folds and/or small edge tears as typical of circulated currency. One has small stains. Please examine photos closely for individual condition details.

From ChatGPT:

5 Vintage Red Seal $2 Bills – 1953–1963 Series – U.S. Notes (Circulated Lot of 5)

Own a tangible piece of mid-century American financial history with this set of five vintage Red Seal $2 bills. Issued by the U.S. Treasury between 1953 and 1963, these are United States Notes (not Federal Reserve Notes), recognizable by their distinctive red seals and serial numbers.

This collection includes five different series, each reflecting changes in Treasury leadership during the era—making it an appealing lot for collectors interested in both currency and historical context.

Included Series & Signatures:

1953 – George M. Humphrey (Secretary), Ivy Baker Priest (Treasurer)

1953A – Robert B. Anderson (Secretary), Ivy Baker Priest (Treasurer)

1953B – C. Douglas Dillon (Secretary), Elizabeth Rudel Smith (Treasurer)

1953C – C. Douglas Dillon (Secretary), Kathryn O’Hay Granahan (Treasurer)

1963 – C. Douglas Dillon (Secretary), Kathryn O’Hay Granahan (Treasurer)

Condition:

Circulated condition with a mix of preservation

2 notes are crisp and relatively clean

3 notes show typical circulation wear, including creases, small corner folds, and/or minor edge tears

One note has light staining

Please review photos carefully for exact condition of each bill

A great starter lot or addition to any U.S. currency collection—especially for those drawn to Red Seal notes and mid-20th century Americana.

And from Claude:

5 Vintage Red Seal $2 Bills | 1953–1963 Series | U.S. Notes | Circulated | Complete Run

Own a tangible piece of American monetary history with this complete 5-bill lot of Red Seal $2 United States Notes, spanning every series issued from 1953 through 1963. Unlike the green-seal Federal Reserve Notes in circulation today, these are United States Notes — a now-defunct currency class issued directly by the U.S. Treasury, making them increasingly scarce and collectible.

This lot is ideal for the collector who wants a complete run of the Red Seal $2 series in a single purchase. Each bill bears a unique combination of Secretary of the Treasury and U.S. Treasurer signatures, reflecting the changing hands of American financial leadership across three presidential administrations.

What’s Included — Series & Signatures:

1953 — Secretary George M. Humphrey / Treasurer Ivy Baker Priest

1953A — Secretary Robert B. Anderson / Treasurer Ivy Baker Priest

1953B — Secretary C. Douglas Dillon / Treasurer Elizabeth Rudel Smith

1953C — Secretary C. Douglas Dillon / Treasurer Kathryn O’Hay Granahan

1963 — Secretary C. Douglas Dillon / Treasurer Kathryn O’Hay Granahan

🔍 Condition:

These bills are circulated and sold as a complete lot. Conditions vary by note: 2 bills are crisp and clean with strong detail, while 3 bills show minor wear consistent with circulation — light creasing, small corner folds, and/or minor edge imperfections. One bill has small staining.

Please review all photos carefully, as they are considered part of the description. What you see is what you get.

Why Collect Red Seal $2 Bills?

The Red Seal $2 United States Note was discontinued after 1963 when the series was replaced by the green-seal Federal Reserve Note version. That makes this complete 5-series run a finite, non-reproducible set — no more will ever be printed. They are legal tender and a beloved staple of mid-century Americana.

Sold as a complete lot of 5. Bills will be shipped securely in protective sleeves. Please see photos for full condition details.

I’m not going to comment, except to say that both titles are too long for eBay. But since you can have multi-turn conversations, you can ask the chatbot to check the title for length and rewrite it.

So how can using chatbots to help write listings benefit vintage sellers?

The biggest boost you will get is that if you ask it to optimize for SEO, the chatbot will do that. Does that sometimes mean the title is clunky? Yes. But if your goal is getting the listing seen and making sales, clunky can be a necessary evil.

You can’t think of everything when you’re writing a listing, particularly buzzwords like aesthetics that are popular. So running a description through a chatbot might give you a few extra sparks.

If you cross-post on more than one site, a chatbot can help you customize the listing for each site. A listing in your own shop can be different from an eBay or Etsy or Poshmark listing. Yes, you can massage the listing for each site, or you can let a chatbot do it for you in a matter of seconds.

Instead of searching the dusty corners of your brain trying to come up with 13 non-repetitive tags for Etsy, it will give you a decent list including known terms people search for and other things you did not think of. There’s no question I’ve improved my tagging skills by learning from the chatbots.

Chatbots can help stave off burnout. When you are writing 10-20 listings a day, day after day, it can be precious hard to bring your A game to every listing. My description for the crocheted dress was accurate but more boring than a slice of white bread without butter or jam. And although the flowery slop the chatbots churned out make me roll my eyes. They did give me something to work with that I could edit to better reflect my personality. Is that cheating? I clearly have decided it is not. But you have to make up your own mind.

When you combine things chatbots can do to streamline your flow with how it can also streamline research (with caveats), using them as tools can be as helpful as using Clorox wipes and Oxiclean soaks-they’re just another thing that helps you do your job better.

How you use chatbots is up to you

It’s up to you if you paste in a listing description directly as a chatbot writes it, or if you use that as a jumping off point.

Chatbots do not have a standard style for listing suggestions and will spit out different formats each time you ask it to produce a listing. For example, sometimes it will add a string of keywords at the bottom of the listing. You decide if you want to use their format.

It’s up to you if you use AI to generate listing photos, including adding models wearing garments being sold.

Chatbots will suggest prices for your listings if you ask it to. You decide if you want to believe it or do your own research.

The exponential growth of AI and the expanding capabilities of chatbots has been explosive. If and how you use it is as personal as what you fill your Ikea bag with when you’re sourcing. The two iron clad rules are to verify everything a chatbot tells you. And to remember that the chatbot is not your friend no matter how many times it tells you your are astute.

IMPORTANT NOTE: We do NOT now, nor will we ever use a chatbot to write an post for this blog. We do use it for research in the same way we use search engines, we verify every piece of information before we include it and we cite sources whenever applicable.

OTHER IMPORTANT NOTE: The use of AI to enhance and alter listing photos is the subject of heated discussion. Using photo editing tools to improve color or eliminate a background is a standard use. Putting your dress on a virtual model or your vase on a virtual table is an emerging trend. It will be interesting to see how that shakes out. I’m old enough to remember the kerfuffle and pearl clutching when National Geographic magazine used early photo editing software to move one of the pyramids for a cover shot in 1982.

There are many controversies surrounding AI including risks to mental health, bias, misinformation, privacy, data usage, legal and ethical considerations and environmental impact. I’m not ignoring these things. They’re too big for a tiny vintage blog to delve into.

How to Spot when Writing is AI: The 6 Elements of AI Style, Huntingthemuse.net

Selling is now even easier with AI, eBay.com