Vinted AI: using ChatGPT and Claude for reselling, and what MCP changes

Ask ChatGPT what price to resell a Carhartt Detroit jacket for on Vinted. You'll get a fluent, well-argued answer with a price range that looks credible. The problem: that range isn't based on a single real sale. The AI didn't measure it, it inferred it from texts it saw during training.

That's the whole paradox of AI for Vinted in 2026. Assistants like ChatGPT, Claude or Gemini have become excellent at writing, structuring and reasoning. But for a reseller, the decisions that matter (what to buy, what price to resell at, which niche to go after) depend on data these assistants don't have.

This article sorts it out: what a general-purpose AI can really do for you, where it gets things wrong, what the Model Context Protocol (MCP) is, explained without jargon, and how the ResellTrack MCP connects your assistant to Vinted sales data.

The essentials in 30 seconds

• ChatGPT and Claude are useful for writing titles and descriptions, structuring a process, and analysing a spreadsheet you give them.
• They don't know Vinted sales. When you ask them for a price, they produce a plausible estimate, not a measurement.
• MCP is an open standard that lets you plug an AI assistant into an external data source, the way you plug a device into a USB-C port.
• The ResellTrack MCP gives your assistant access to the Vinted sales data collected by ResellTrack: sold prices, sellers, brands, niches, trends.
• The AI analyses, you decide. Check the numbers that matter and don't automate any action on your Vinted account.

What a general-purpose AI can do for a Vinted reseller

Before talking about limits, let's be fair: an AI assistant saves a lot of time on anything involving text and structure. It's actually the most common use when people search for "ChatGPT Vinted": writing listings.

Writing and optimising listings

Give the AI the brand, model, size, condition and any flaws, and it suggests a clean title and description in seconds. With a stock of 50 items or more, the time saved is considerable.

• Structured titles that match how buyers search: brand, item type, colour, size.
• Consistent descriptions across your whole stock, with the same sections (condition, measurements, material).
• Reply templates for buyers: measurements, bundles, counter-offers.

To go further on what makes a listing sell, read our article on how the Vinted algorithm works: a good title is useless if the listing is badly positioned.

Structuring your business

This is an underrated use. An assistant is very good at:

• building a stock and margin tracking spreadsheet;
• calculating a net margin from your buying price, your fees and your resale price;
• analysing an export you give it (your own sales, for example) and drawing conclusions from it.

What all these uses have in common: you provide the information. The AI formats it or crunches it. As soon as the question is about the market itself, things change.

What a general-purpose AI can't do (and why it makes things up)

It doesn't have Vinted sales data

A model like ChatGPT or Claude is trained on large amounts of public text, up to a cut-off date. It has no access to the history of items sold on Vinted, nor to the prices they went for, nor to how long they took to sell. That information isn't in its training data in any usable form.

Even when the assistant has web search, it mostly finds live listings, in other words asking prices. And a listed price isn't a sold price. A listing at €80 can stay online for three months; the one at €45 sells in two days. It's the second one you care about.

It produces plausible answers, not verified ones

A language model generates the most likely continuation of a text. When it doesn't know, it doesn't always say "I don't know": it produces an answer that looks like a good answer. That's what's called a hallucination.

In practice, for a reseller, this means:

• Made-up prices: a "realistic" range that can be a long way from the current market.
• Outdated trends: the AI cites brands that "work" based on old blog posts, not on what's selling this month.
• Generic niches: "vintage", "sneakers", "luxury". Answers everyone gets, so markets that are already saturated.

If you buy stock based on those answers, you're making a financial decision on data that doesn't exist. That's exactly the mistake we describe in our guide to analysing a Vinted niche's profitability: profitability is measured on sales, not on a hunch.

It doesn't know the context of a specific seller or listing

Paste a listing URL into ChatGPT: it won't know how many similar items have sold, at what median price, or whether this seller shifts stock quickly. It's missing the basis for comparison.

The Model Context Protocol (MCP) explained simply

That's where MCP comes in. The term shows up more and more in searches like "Vinted MCP" or "Claude Vinted", often without a clear explanation.

Definition

The Model Context Protocol is an open-source standard introduced by Anthropic (the company behind Claude) in November 2024. Its official documentation defines it as a standard for connecting AI applications to external systems: data sources, tools, workflows.

The analogy used by the project itself says it well: MCP is a USB-C port for AI applications. Just as a USB-C port lets you plug in any compatible device, MCP lets you plug any compatible assistant into any compatible data source.

How it works, without the jargon

There are two sides:

• The client: your AI assistant (Claude, and other MCP-compatible assistants such as ChatGPT, which supports it too).
• The server: a service that exposes data and tools. For example, a server that can answer "give me the sales for this brand in this category".

When you ask a question, the assistant realises it's missing some information, calls the right tool on the server, receives real data, then writes its answer from that data. You keep writing in plain English; the assistant takes care of querying the source.

The difference is fundamental: the AI no longer guesses, it looks things up.

Not all Vinted MCPs are equal

Several Vinted MCP servers already exist, including open-source ones. Most of them query public live listings: item search, asking prices, profiles. That's useful for exploring supply, but it runs into the limit mentioned above: a live listing tells you what sellers hope for, not what buyers pay.

To decide on a purchase or a price, the data that counts is the recorded sale: which item sold, at what price, and how quickly. That's precisely the data ResellTrack collects and consolidates.

The ResellTrack MCP: plugging your AI into Vinted sales

The ResellTrack MCP connects Claude, or any other MCP-compatible AI assistant, directly to ResellTrack's Vinted sales data. Your assistant keeps its writing and reasoning skills, and gains what it was missing: a database of real sales to query.

What the server actually exposes

• Resale price estimates from a Vinted listing URL.
• Seller analysis: their sales and how they break down by niche.
• Brand and niche statistics: volumes, prices, speed of sale.
• Niche trends: what's speeding up or slowing down.
• An item's price history.
• Live supply: the active listings on a segment, to measure competition.
• Sold item search.
• Image search: finding comparable items from a photo.

All amounts are in euros. Data is consolidated with a lag of up to 24 hours: it's a reliable basis for analysis, not an instant feed.

These capabilities mirror what you find in the app (Trending, Top Sellers, Median sold price, Scan, Lens), but available in a conversation. Each feature is detailed on the ResellTrack features page.

Example questions to ask

The point of MCP is to phrase your questions the way you'd put them to an analyst. A few examples:

• "Which men's jacket brands are selling fastest this month, and at what median price?"
• "Estimate the resale price of this listing: [Vinted listing URL]."
• "Analyse this seller: how many sales, in which niches, at what average price?"
• "Which niches are growing in the women's shirts category?"
• "How many listings are currently live for this brand in this category, compared with sales volume?"
• "Here's a photo of a piece I found in a charity shop: which comparable items have sold, and at what price?"

What an answer looks like

For illustration, here's the structure of a typical answer to the first question (the points below describe the format, not real results):

• a ranking of brands by number of sales over the period;
• for each one, the median sold price and a speed-of-sale indicator;
• the volume of active listings on the other side, to spot segments where supply is still low;
• a summary written by the assistant, highlighting the most interesting brand and category combinations.

The difference with an unconnected AI lies in where the numbers come from: they come from recorded sales, not from a generated estimate.

Three concrete use cases for a reseller

1. Sourcing: knowing what to buy before you buy

Sourcing is the stage where mistakes cost the most: a bad purchase ties up capital for weeks. With MCP, you can validate a lead before spending a single euro.

• Ask for the brands that sell best in a given category.
• Cross-check with live supply to avoid saturated segments.
• In a charity shop or at a car boot sale, send a photo to find comparable sales and set your maximum buying price.

This approach is covered in detail in our guide on how to identify a profitable Vinted niche.

2. Pricing: neither too high nor too low

Paste the URL of your listing or a comparable one and ask for a resale price estimate. You get a benchmark based on real sales, which you can then adjust for the condition and rarity of your piece.

3. Competitive intelligence: understanding what works for others

Ask your assistant to analyse a seller you often come across in your niches: sales volume, split by niche, price level. In a few exchanges, you understand their positioning and the segments they occupy. You can also spot the most active sellers in a niche to know who you're up against.

For a complete method inside the app itself, see our tutorial on how to use ResellTrack to find Vinted niches.

Connecting ResellTrack to your AI assistant

You connect from your assistant, not from Vinted. The principle is the same for every MCP-compatible assistant:

1. Open your AI assistant's connector settings (the section that manages MCP connectors or integrations).
2. Add the ResellTrack connector.
3. Sign in with your ResellTrack account when the assistant asks you to.
4. Ask a simple first question to check the assistant is really using ResellTrack data, for example a price estimate on a listing you know.

The exact labels vary from one assistant to another; check its help pages if needed.

The AI connection is available from the Pro plan (€34/month, no commitment). The Starter, Pro and Business plans are detailed on the ResellTrack pricing page.

Limits and good practice

Plugging an AI into real data greatly reduces the risk of error. It doesn't eliminate it. A few simple rules.

Check the numbers that involve money

The assistant interprets the data it receives, and an interpretation can be off: wrong category, misunderstood period, overly quick summary. Before a large stock purchase, open the key figures in the app or ask the assistant to spell out which data it's relying on.

Ask precise questions

"What sells well?" gets you a vague answer. "Which women's jumper brands have sold fastest over the last 30 days, with the median price and the number of live listings?" gets you an answer you can act on. Always specify the category, the gender, the period and the metric you want.

Keep the data lag in mind

Data is consolidated within up to 24 hours. That's more than enough to decide on sourcing or a price, but don't expect to see the sale that happened ten minutes ago.

Don't automate any action on your Vinted account

The ResellTrack MCP is an analysis tool: it reads market data, it doesn't act on your account. That's deliberate. Automating actions on Vinted (listing, follow-ups, sending messages, buying) through a script or an AI agent exposes you to account restrictions. We cover the risks in our article on Vinted bots.

The right split: the AI analyses and prepares, you list and you decide.

Connected AI, app or extension: which should you choose?

They're not competing tools, they're three ways to access the same data:

• The ResellTrack app: for visual exploration, the instant Scan of a niche (a Score from 0 to 100 without creating a tracker or waiting), the Trending page and tracking over time.
• The Chrome extension: for getting the data directly on Vinted, on a seller profile or an item page. See our article on the ResellTrack Vinted extension.
• The MCP: for asking cross-cutting questions in plain English and getting a written summary, without jumping between screens.

To see where ResellTrack sits among the other solutions on the market, check our comparison of the best Vinted reselling tools in 2026.

Conclusion

Artificial intelligence applied to Vinted reselling is only useful if it rests on the right data. A general-purpose AI writes your listings better and faster than you, but it makes things up as soon as you talk to it about prices, brands or niches.

The Model Context Protocol fixes that by plugging your assistant into a reliable source. With the ResellTrack MCP, Claude or your compatible assistant queries real Vinted sales instead of guessing. You keep the decision, with a solid basis for making it.

Frequently asked questions

Can ChatGPT estimate the price of a Vinted item?

On its own, not reliably. ChatGPT has no access to Vinted's sales history: it produces a plausible range from its general knowledge, or relies on live listings, which reflect asking prices rather than sold prices. Connected to a sales source via MCP, however, it can base its estimate on recorded sales.

What is a Vinted MCP?

It's a server based on the Model Context Protocol that lets an AI assistant query Vinted-related data. Depending on the server, that data can be live listings or, in ResellTrack's case, consolidated Vinted sales: sold prices, sellers, brands, niches and trends.

Does the ResellTrack MCP only work with Claude?

No. It works with Claude and with other MCP-compatible AI assistants. You connect from the assistant's connector settings, with your ResellTrack account.

Which plan do I need to connect my AI to ResellTrack?

The AI connection is available from the Pro plan, at €34 a month with no commitment, with 100 requests a month. The Business plan at €69 a month also includes it, with 500 requests a month. The Starter plan, at €24 a month, doesn't include the AI connection.

Can AI manage my Vinted account for me?

That's not the ResellTrack MCP's job: it's limited to analysing market data and doesn't act on your account. Automating actions on Vinted (listings, messages, purchases) exposes your account to restrictions. Use AI to analyse and prepare, and keep your actions on Vinted manual.